an impact evaluation of india’s second and
TRANSCRIPT
An Impact Evaluation of India’s Second and Third Andhra PradeshIrrigation ProjectsA Case of Poverty Reduction with Low Economic Returns
An Impact Evaluation of India’s Second and Third Andhra PradeshIrrigation ProjectsA Case of Poverty Reduction with Low Economic Returns
THE WORLD BANK
45406
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
2006 Annual Report on Operations Evaluation
Annual Review of Development Effectiveness 2006: Getting Results
Addressing the Challenges of Globalization: An Independent Evaluation of the World Bank’s Approach to Global Programs
Assessing World Bank Support for Trade, 1987–2004: An IEG Evaluation
Books, Buildings, and Learning Outcomes: An Impact Evaluation of World Bank Support to Basic Education in Ghana
Brazil: Forging a Strategic Partnership for Results—An OED Evaluation of World Bank Assistance
Bridging Troubled Waters: Assessing the World Bank Water Resources Strategy
Capacity Building in Africa: An OED Evaluation of World Bank Support
China: An Evaluation of World Bank Assistance
The CGIAR at 31: An Independent Meta-Evaluation of the Consultative Group on International Agricultural Research
Committing to Results: Improving the Effectiveness of HIV/AIDS Assistance—An OED Evaluation of the World Bank’sAssistance for HIV/AIDS Control
Country Assistance Evaluation Retrospective: OED Self-Evaluation
Debt Relief for the Poorest: An Evaluation Update of the HIPC Initiative
A Decade of Action in Transport: An Evaluation of World Bank Assistance to the Transport Sector, 1995–2005
The Development Potential of Regional Programs: An Evaluation of World Bank Support of Multicountry Operations
Development Results in Middle-Income Countries: An Evaluation of the World Bank’s Support
Economies in Transition: An OED Evaluation of World Bank Assistance
Engaging with Fragile States: An IEG Review of World Bank Support to Low-Income Countries Under Stress
The Effectiveness of World Bank Support for Community-Based and -Driven Development: An OED Evaluation
Evaluating a Decade of World Bank Gender Policy: 1990–99
Evaluation of World Bank Assistance to Pacific Member Countries, 1992–2002
Extractive Industries and Sustainable Development: An Evaluation of World Bank Group Experience
Financial Sector Assessment Program: IEG Review of the Joint World Bank and IMF Initiative
From Schooling Access to Learning Outcomes: An Unfinished Agenda—An Evaluation of World Bank Support to PrimaryEducation
Hazards of Nature, Risks to Development: An IEG Evaluation of World Bank Assistance for Natural Disasters
How to Build M&E Systems to Support Better Government
IEG Review of World Bank Assistance for Financial Sector Reform
Improving Investment Climates: An Evaluation of World Bank Group Assistance
Improving the Lives of the Poor Through Investment in Cities
Improving the World Bank’s Development Assistance: What Does Evaluation Show?
Maintaining Momentum to 2015: An Impact Evaluation of Interventions to Improve Maternal and Child Health and NutritionOutcomes in Bangladesh
New Renewable Energy: A Review of the World Bank’s Assistance
Pakistan: An Evaluation of the World Bank’s Assistance
Pension Reform and the Development of Pension Systems: An Evaluation of World Bank Assistance
Poland Country Assistance Review: Partnership in a Transition Economy
The Poverty Reduction Strategy Initiative: An Independent Evaluation of the World Bank’s Support Through 2003
The Poverty Reduction Strategy Initiative: Findings from 10 Country Case Studies of World Bank and IMF Support
Power for Development: A Review of the World Bank Group’s Experience with Private Participation in the Electricity Sector
Putting Social Development to Work for the Poor: An OED Review of World Bank Activities
Small States: Making the Most of Development Assistance—A Synthesis of World Bank Findings
Social Funds: Assessing Effectiveness
Sourcebook for Evaluating Global and Regional Partnership Programs
Water Management in Agriculture: Ten Years of World Bank Assistance, 1994–2004
World Bank Assistance to the Financial Sector: A Synthesis of IEG Evaluations
The World Bank in Turkey: 1993–2004—An IEG Country Assistance Evaluation
World Bank Lending for Lines of Credit: An IEG Evaluation
All IEG evaluations are available, in whole or in part, in languages other than English. For our multilingual section, please visit
http://www.worldbank.org/ieg
RECENT IEG PUBLICATIONS
An Impact Evaluation of India’s
Second and Third Andhra Pradesh
Irrigation Projects
A Case of Poverty Reduction with
Low Economic Returns
W O R L D B A N K I N D E P E N D E N T E V A L U A T I O N G R O U P
2008
The World Bank
Washington, D.C.http://www.worldbank.org/ieg
©2008 The International Bank for Reconstruction and Development / The World Bank
1818 H Street NW
Washington DC 20433
Telephone: 202-473-1000
Internet: www.worldbank.org
E-mail: [email protected]
All rights reserved
1 2 3 4 11 10 09 08
This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The
findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive
Directors of The World Bank or the governments they represent.
The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors,
denominations, and other information shown on any map in this work do not imply any judgement on the part of The
World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries.
Rights and Permissions
The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission
may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank
encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly.
For permission to photocopy or reprint any part of this work, please send a request with complete information to the
Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-
4470; Internet: www.copyright.com.
All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher,
The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: [email protected].
Cover photo: Construction of a main canal in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.
ISBN-13: 978-0-8213-7542-6
e-ISBN-13: 978-0-8213-7543-3
DOI: 10.1596/978-0-8213-7542-6
Library of Congress Cataloging-in-Publication Data have been applied for.
World Bank InfoShop Independent Evaluation Group
E-mail: [email protected] Knowledge Programs and Evaluation Capacity
Telephone: 202-458-5454 Development (IEGKE)
Facsimile: 202-522-1500 E-mail: [email protected]
Telephone: 202-458-4497
Facsimile: 202-522-3125
Printed on Recycled Paper
Contents
i i i
vii Abbreviations
ix Acknowledgments
xi Foreword
xiii Preface
xv Executive Summary
1 1 Introduction3 The Irrigation Paradox6 The Andhra Pradesh Context7 Evaluation Questions7 Study Approach and Overview9 Overview of the Report
11 2 The Projects13 Second Andhra Pradesh Irrigation Project14 Third Andhra Pradesh Irrigation Project 14 Project Implementation16 Summary
19 3 Water User Associations21 Origins of PIM in Andhra Pradesh21 The Current Wave of PIM22 WUA Membership24 What Do WUAs Do and Who Does It?27 Summary
29 4 Economic Benefits31 Direct Production Impacts35 Direct Income Effects: Higher Farm Income36 Indirect Income Effects: Employment Generation38 Who Benefits? Poverty Reduction and Distributional Effects of New Irrigation40 Revisiting the ERR43 Resolving the Irrigation Paradox
45 5 Summary and Lessons Learned47 Findings on the Six Evaluation Questions48 On Selecting Irrigation Investments49 On Participatory Irrigation Management49 On the Impact of Irrigation
51 Appendixes53 A: Sample Selection of IEG Irrigation Survey57 B: Regression Results67 C: Single- and Double-Difference Impact Estimates81 D: Data Tables93 E: The Farm Model, Rate of Return Analysis, and Distributional Impacts
107 F: Village Profiles109 G: Comments from Government of Andhra Pradesh 121 H: Impact Evaluation
123 Endnotes
129 Bibliography
Boxes5 1.1 Irrigation: The Basics7 1.2 Some Local Terminology8 1.3 The IEG Survey
25 3.1 Diversion of Water to Tanks Limits the Number of Beneficiaries43 4.1 Improving the Quality of Sensitivity Analysis in Bank Appraisals
Figures15 2.1 Irrigated Area in SRSP 16 2.2 Irrigation Sources, 2005 and 200622 3.1 Households Accessing Canal and Tank Irrigation Systems Most Likely to
Be in a WUA23 3.2 Likelihood of WUA Membership Increases Sharply with Wealth23 3.3 Large Landowners Most Likely to Be WUA Members; Marginal Farmers
also Active24 3.4 Lorenz Curve for Land Ownership by Irrigation Source26 3.5 Larger Landholders More Likely to Think Water Is Fairly Distributed, as
Are WUA Members32 4.1 Irrigation Increases Yields . . . But by Less than Assumed at Appraisal 33 4.2 Cropping Pattern by Irrigation Source, 2006 35 4.3 Cropping Intensity by Number of Irrigation Sources, 200635 4.4 Increased Cropping Intensity Comes from Additional Growing Seasons38 4.5 Seasonal Employment Effects39 4.6 Distribution of Irrigation Benefits Skewed toward the Better Off40 4.7 Poverty Impact of Irrigation
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
i v
41 4.8 Income Distribution Improved between 2005 and 200643 4.9 Investments in SRBC Had No Chance of Providing an Acceptable Rate of
Return
Tables6 1.1 Bank-Supported Irrigation Projects in Andhra Pradesh
25 3.1 Participation in WUA Activities, 200636 4.1 Crop Failure by Irrigation Source 41 4.2 Estimates of the Economic Rate of Return42 4.3 Alternative Estimates of the Economic Rate of Return
C O N T E N T S
v
v i i
Abbreviations
AP II Second Andhra Pradesh Irrigation Project
AP III Third Andhra Pradesh Irrigation Project
APERP Andhra Pradesh Economic Restructuring Project
APFMIS Andhra Pradesh Farmers Management of Irrigation Systems (Act)
CAD Command area development
CADA Command Area Development Authority
Cusecs Cubic meters a second
DC Distributary committee
ERR Economic rate of return
ha Hectares
I&CADD Irrigation and Command Area Development Department
ICR Implementation Completion Report
IEG Independent Evaluation Group
IMT Irrigation Management Transfer
O&M Operations and maintenance
PIM Participatory irrigation management
Rs Rupees
RLP Rural Livelihoods Project
SAR Staff Appraisal Report
SRBC Srisailam Right Branch Canal
SRSP Sriramasager Project
WOP Without project
WP With project
WUA Water user association
Watering a crop from a field canal in India. Photo by Ray Witlin, courtesy of World Bank Photo Library.
i x
Acknowledgments
The study was made possible by the cooperation
of the government of Andhra Pradesh. The In-
dependent Evaluation Group (IEG) is particu-
larly indebted to Dr. C.V. S. K. Sarma (IAS, Former
Principal Secretary, I&CAD Department, govern-
ment of Andhra Pradesh) and Mr. A. Subba Rao
(Chief Engineer, Sriramasagar Project, I&CAD De-
partment, government of Andhra Pradesh), as
well as many irrigation officials in Warangal. The
staff of the Agricultural Finance Corporation not
only organized the field work but also provided
logistical support to the IEG team in all aspects of
its work. From the Agricultural Finance Corpora-
tion, particular thanks are due to the general man-
ager and project leader for this study, Dr. D. V.
Ramana, to Mr. P. H. Khan (field organizer), Mr. B.
Rajaramakrishna (field supervisor and computer
programmer), and Mr. K. K. Rao (project advisor
and field supervisor); and to field supervisors
Mr. Y. Seshu Ganesh, Mr. C. Manohara, Mr. A.C.
Rao, and Mr. Srinivas Rao. Thanks are also due to
the villagers who took time to reply to question-
naires or to meet with the study team. From the
World Bank, R.S. Pathak has been a useful source
of information and guidance.
This report was prepared by Howard White (task
manager) and Edoardo Masset (consultant), with
inputs from Nina Blöndal (consultant), under the
guidance of Alain Barbu. Useful comments have
been received from Keith Oblitas and Keith Pitman
and the external reviewers Peter Hazell, Michael
Lipton, and Richard Palmer-Jones. The report was
edited by William Hurlbut and Heather Ditt-
brenner. Administrative support was provided by
Soon-Won Pak.
Director-General, Evaluation: Vinod Thomas
Director, Independent Evaluation Group–World Bank: Cheryl Grey
Manager, Sector, Thematic, and Global Evaluation: Alain Barbu
Task Manager: Howard White
Workers digging irrigation ditch in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.
x i
Foreword
Irrigation has made a major contribution to
poverty reduction in the past decades, enabling
higher yields and better nutrition. Despite these
achievements, large-scale irrigation schemes have
usually yielded low returns and attracted negative
publicity because of their adverse environmental
and social impacts. As a result, the Bank has largely
switched its support for irrigation away from new
construction toward rehabilitation and policy re-
form. This evaluation supports the need for reform
but shows that there are substantial benefits from
further investment in infrastructure.
This study analyzes these issues through an im-
pact evaluation of one of the last “old genera-
tion” of projects in which the Bank directly
supported creation of a new irrigation scheme:
India’s Second and Third Andhra Pradesh Irriga-
tion Projects (AP II and AP III). Together these proj-
ects created a new command area, the Srisailem
Right Branch Canal (SRBC), and rehabilitated an
existing one that had been constructed with Bank
assistance, the Sriramasagar Project.
The impact analysis, which uses a quasi-
experimental design and is based on new survey
data and reanalysis of existing data, confirms that
irrigation has a substantial poverty-reducing im-
pact: in the first year of receiving canal irrigation
the poverty rate falls by a quarter. Although larger
land owners benefit the most in absolute terms,
the greatest income growth accrues further down
the income distribution, mainly because of the
growth in wage employment. The long-run
poverty-reducing effects are greater still, as house-
holds become less vulnerable to rainfall fluctua-
tions and so escape the poverty trap of periodic
shocks undermining their asset base.
But the returns to investments anticipated at ap-
praisal did not materialize: Yield increases were
lower than forecast and the expected diversifica-
tion did not occur; indeed, canal-irrigated farms
have been less diversified than those without ir-
rigation. The expectations at appraisal were overly
optimistic on both yield and diversification. The
return was further undermined for AP II and AP
III by long delays and large cost overruns. These
problems were most severe in the case of the
SRBC project, for which the first water was re-
ceived by farmers nearly 20 years after construc-
tion started—resulting in a rate of return of, at
best, zero.
The return to the rehabilitation component was
higher but was still less than 10 percent. In sum-
mary, the investments were efficacious but not ef-
ficient. Investments in large-scale construction
need to establish mechanisms to ensure that con-
struction is completed on time and on budget—
the government of Andhra Pradesh is now
adopting such measures. Part of avoiding delays
is ensuring adequate preparation, which did not
happen in this case, partly because the Bank
raised new design issues once implementation had
begun.
Andhra Pradesh has been at the forefront of par-
ticipatory irrigation management through water
user associations (WUAs). More than 10,000 WUAs
were created across the state, with elections of of-
ficers being held in a single day. These WUAs have
played an important role in improving the qual-
ity of rehabilitation works.
But WUAs cannot take full responsibility for op-
erations and maintenance (O&M); they often lack
the technical capacity to do so. They also lack the
financial resources, as funds collected through the
water cess are insufficient for adequate O&M, es-
pecially as collection rates remain low. In part
this is because the problem of inadequate water
at the tail ends has not been solved. WUAs are not
able to solve problems of water allocation, espe-
cially those between communities. Excess water
use in the head reaches, including sabotage of
water control measures, remains prevalent. There
will continue to be a role for government in water
management, and further means of improving
water allocation need to be considered.
Although the Independent Evaluation Group has
engaged with the government of Andhra Pradesh
throughout the study, the government does not
share all the conclusions of this evaluation. In
particular, it has pointed out that it has taken
steps to rectify the construction inefficiencies
that plagued these projects. The government has
also stressed its continued commitment to de-
veloping WUAs, which it believes will overcome
what it sees as “teething problems” in institu-
tional development.
Among the key findings of the study are the
following:
• Large-scale irrigation can—and in this case did—
have a significant poverty-reduction impact.
• Large-scale construction yields low returns
when there are cost overruns and delays. These
can be avoided with more rationalized selection
and planning and appropriate design and pro-
curement procedures.
• WUAs can help improve water management
but are not a panacea; there is a continued
role for government in both technical aspects
and facilitating equitable water distribution.
x i i
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Vinod Thomas
Director-General, Evaluation
x i i i
Preface
The Independent Evaluation Group (IEG) of the
World Bank has undertaken impact evaluations of
the Bank’s support to irrigation in Andhra Pradesh,
India (under AP Irrigation II and III), and of the
U.K. Department for International Development-
supported Rural Livelihoods Project (RLP).
This is one of a series of IEG impact evaluations
(see appendix H). IEG’s program of impact eval-
uation is in part carried out under a Department
for International Development–IEG partnership
agreement; hence the focus on RLP. However,
survey villages are also covered by the Bank-
supported DPIP project, so that the findings are
also relevant to this project.
For this purpose, a first-round survey was con-
ducted in June 2005 in the five RLP districts and
the Warangal district, an area of recent irrigation
expansion along the Kakatiya canal under the Sri-
ramasagar Project, which was rehabilitated with
World Bank support. A second-round survey was
undertaken in Warangal in March 2006, along
with qualitative data collection. This report pres-
ents an analysis of the economic and social impact
of the two irrigation projects.
The report is the first IEG will publish concern-
ing rural development in Andhra Pradesh. Sub-
sequent reports will review the impact of RLP
following a third survey round in June 2007 and
compare the effectiveness of different interven-
tions for reducing rural poverty. This next study
will explicitly compare the effectiveness of large-
scale irrigation investments, the subject of this
study, with the smaller-scale groundwater-based
interventions promoted by RLP, and it will review
the empowerment route to poverty reduction
embodied in both RLP and Indira Kranti Patham.
The findings presented in this report comple-
ment those in the recent IEG study Water Man-
agement in Agriculture: Ten Years of World Bank
Assistance, 1994–2004 (IEG 2006b). Many of the
general lessons in that report concerning prob-
lems encountered in irrigation investments are ex-
emplified by the two projects considered in this
report. Building on the sector review, this report
presents new evidence on the strengths and weak-
nesses of water user associations, which are an im-
portant part of the Bank’s current approach to
irrigation. This study also goes into greater detail
than a typical IEG project assessment on social and
economic impacts, using survey data to identify
the distribution of both direct and indirect ben-
efits from irrigation investments.
A draft of the report was discussed with officials
of the government of Andhra Pradesh in Hyder-
abad in March 2007, and written comments were
subsequently received (attached as appendix G,
with an IEG response to the major points raised).
These comments have been taken into account
as appropriate in the final version of this report.
Building a main brick canal in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.
x v
Executive Summary
For many developing countries, the productivity
of irrigation investments has been falling as a re-
sult of poor operations and maintenance (O&M).
This problem has increasingly been addressed in
development projects by putting users and com-
munities in charge of O&M, an approach known
as participatory irrigation management (PIM).
However, the effect of this approach on equity, and
hence its implications for poverty reduction, have
yet to be fully explored.
The Independent Evaluation Group (IEG) of the
World Bank has examined the paradoxical nature
of irrigation projects by evaluating the impact of
two such projects in the Indian state of Andhra
Pradesh. Irrigation Projects II and III (AP II and AP
III) aimed to irrigate 393,000 hectares (ha) of
land—65,000 ha served by the Srisailam Right
Bank Canal (SRBC), 163,000 ha served by a re-
habilitated canal, and 165,000 ha gained from ex-
tending the system under the Sriramasagar Project
(SRSP). Together these projects were expected to
benefit more than 300,000 farm households.
The evaluation findings have implications for
the management of irrigation projects. Al-
though the World Bank has largely disengaged
from financing new irrigation construction in
India, the findings of the evaluation are of inter-
est, as the government of Andhra Pradesh is plan-
ning major investments in new irrigation schemes.
The findings are also relevant for community-
driven development and other participatory ap-
proaches and to implementation of other reforms
such as water pricing, both of which are part of
the Bank’s current irrigation agenda in India. In-
deed, Bank finance for AP III was originally justi-
fied based on the support it provided to the
reform process in Andhra Pradesh.
Andhra Pradesh has been a leader in the use
of PIM since the late 1990s, when PIM was piloted
in the SRSP. Previously, irrigation O&M was the re-
sponsibility of government agents whose salaries
consumed so much of the O&M budget that lit-
tle went to O&M itself. In 1997, an act of Parlia-
ment handed responsibility for water management
to water user associations (WUAs), which were to
assist in collecting water charges and were to re-
ceive a large share of this revenue to undertake
O&M. The act also provided for the establish-
ment of distributary committees and higher-level
project committees to help manage the O&M.
The World Bank has been closely involved in
these reforms, financing WUA training and irri-
gation system rehabilitation undertaken with the
involvement of WUAs.
Economic ImpactsThe economic impacts of the irrigation proj-
ects were undermined by problems com-
monly encountered in large civil works
programs: cost overruns and construction
The widespread expansion of irrigated farming has largely removed
the threat of malnutrition and premature death for millions around the
world and has had demonstrated poverty-reduction effects. Nonethe-
less, large irrigation projects have fallen out of favor, criticized for their cost-
liness, their environmental impacts, the resettlement problems they cause, and
particularly their poor sustainability record.
delays. AP II did not complete any irrigation
works and thus produced no benefits whatsoever.
AP III completed the investments started under AP
II. Although work started in 1988, no farmers were
receiving water in the SRBC area when AP III
closed in 2004. The planned area will be irrigated
by 2007, nearly 20 years after the project started.
Such long delays make the likelihood of an ac-
ceptable return on investments very remote.
There has been a positive impact on both
crop yields and cropping intensity, but the
expected diversification into high-value
crops has not occurred. An IEG household
survey in a newly irrigated area of the SRSP in the
Warangal district found that irrigated farms are less
diversified than those that are not irrigated. This
lack of diversification is confirmed by other data
sources for different parts of the command area.
This is partly because the water pricing structure
does not discourage excessive use and wastage,
which means that, contrary to official policy, irri-
gated farms grow crops (mostly paddy) requiring
large volumes of water, depriving farms in lower
reaches of water. With the benefit of hindsight, the
expectations at appraisal were overly optimistic
on both yield and diversification.
The bulk of the direct benefits from higher
farm income accrues to the top quarter of
beneficiaries. Indirect benefits from higher em-
ployment are spread more evenly across the in-
come distribution, although the poorest receive
fewest benefits. These indirect benefits in partic-
ular cause irrigation to have a substantial one-off
impact on poverty reduction (on the order of a 10
percentage point reduction in the poverty head-
count as a result of irrigation reaching a village)
and a small progressive impact on distribution. This
is because the greatest income growth from
irrigation is experienced by the second
wealth quartile, a situation that allows a
substantial impact on poverty despite the
skewed distribution of absolute benefits.
The longer-run impact on poverty will be
greater than the immediate impact, because
irrigation ends the cycle of negative income
and debt that undermines asset accumula-
tion. One of the chief benefits of irrigation is the
smoothing effect that it has on incomes, reduc-
ing year-to-year fluctuations in production caused
by variations in rainfall.
The beneficial impacts are less than ex-
pected, as the projects suffered from overly
optimistic appraisals. Yield increases average
about half of appraisal estimates. Hence, given the
other problems encountered by these projects,
the overall return on the investment in SRBC
and SRSP is just under 2 percent (compared with
the appraisal estimate of 24 percent). Even al-
lowing for more optimistic assumptions about
future yields, the economic rate of return does
not rise above 5 percent. Thus, the projects are
a large subsidy to farmers, who do indeed ben-
efit. Moreover, the distribution of benefits is
skewed toward the better off. Herein lies the ex-
planation for the falloff of support for irrigation
despite its poverty-reducing effects: poverty re-
duction is achieved by the large transfer involved
in irrigation financing.
This raises two questions. First, can large-scale ir-
rigation schemes be self-financing and still re-
duce poverty? Second, is subsidizing large-scale
irrigation the best use of a subsidy to reduce rural
poverty? This study answers only the first ques-
tion: schemes can both be self-financing and re-
duce poverty only if construction is efficiently
implemented. Once a scheme is functioning, the
benefits it provides (in good years) are more than
sufficient to pay for the system’s recurrent costs.
These conclusions support the Bank’s shift
to focusing on institutional issues in irri-
gation, including water pricing. The second
question will be the subject of a later IEG study
that considers alternative investments in rural
areas of Andhra Pradesh.
The economic impact findings from this study
are similar to those of other IEG evaluations in the
irrigation sector, including the recent sector re-
view (IEG 2006b): irrigation project outcomes
are restrained by the failure to meet overly
ambitious construction schedules, and in-
x v i
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
vestment viability relies on overly optimistic
estimates of benefits.
The expected economic rate of return at appraisal
is often far above the ex post rate, which has
rarely been above the 10 percent threshold. The
Bank has responded to these problems in Andhra
Pradesh, as it has elsewhere, by engaging in the
reform agenda, one focus of which has been the
creation of WUAs. Far less progress has been
made on establishing a pricing policy that en-
courages optimal water usage. The emphasis on
WUAs is consistent with the Bank’s emphasis on
community involvement. But a growing body
of evidence, including this report, suggests
that community groups are unlikely to live
up to the hopes invested in them.
Social ImpactsAndhra Pradesh has been a leader in pro-
moting WUAs, an approach that was nur-
tured under AP III. But the effectiveness of
WUAs has not met expectations, and there
is a need for a continuing government role
if the irrigation system is to run efficiently.
The expansion and durability of WUAs has
been influenced by political factors. Fol-
lowing the 1997 act of Parliament, more than
10,000 WUAs were created across Andhra Pradesh.
This occurred in part because the chief minister
at the time had a particular interest in promoting
grassroots structures. However, WUAs ceased to
function when elections were not held in 2002 at
the end of WUA officials’ first term of office. The
planned election of management committees at
the project level in irrigation schemes was shelved
and has still not taken place, although elections
are still planned.
WUA membership is not representative of
the population served. According to the 1997
act, all cultivating households are WUA members.
However, an IEG household survey in 18 villages
in the Warangal district found that half of house-
holds with access to canal or tank irrigation were
members, but just 19 percent of all cultivating
households in the irrigation scheme areas were
members. Membership was strongly skewed to-
ward the better off, and leadership of WUAs was
dominated by members of higher castes (notably
the president, who often runs the WUA with little
consultation). Although supporters of PIM claim
it will benefit the poor, the landless—who are
the poorest—are automatically excluded. Even
among cultivators, it is the better off who are
most likely to participate, partly because they
have better access to irrigated plots.
A positive feature of irrigation development
since the 1997 reforms has been rehabili-
tation of sections of the canal network:
water now reaches more areas. WUAs were
involved in the identification of these works and
undertook most of the rehabilitation. This in-
volvement has improved the effectiveness and
efficiency of these works, even though WUA pres-
idents often did not consult other members and
benefited personally as contractors. The sustain-
ability of the system, however, is open to question.
Even though higher tariffs have been imposed, the
water charges cannot fully finance the amounts
needed for O&M. And with low collection rates,
charges currently fall far short of costs. The re-
forms have not broken the vicious circle whereby
many farmers will not pay for irrigation services
because these services are poor; many others
who do receive water do not pay either.
The farmers’ committee structure (WUAs,
distributary committees, and project com-
mittees) is intended to ensure equitable
allocation of water but has limited capacity
to do this effectively, especially at higher
levels, as the structure is incomplete. The cre-
ation of WUAs and higher-level distributary com-
mittees has not yet been able to prevent the
age-old problem of farmers in the head reaches
of an irrigation system taking more water than in-
tended, thus depriving farmers in the tail reaches.
This problem might have been addressed by proj-
ect committees, but they were never formed, so
the management structure was left incomplete.
None of the tail-reach villages covered by the IEG
study has received irrigation water. Within villages,
canal water is frequently diverted to an existing
E X E C U T I V E S U M M A R Y
x v i i
tank, so farmers drawing water from that tank,
who have their own WUA, are the ones who ben-
efit. Breaches of the canal to take water are also
common, and again WUAs seem powerless to
act. Thus far, most WUAs seem unable to break
the existing pattern of social and political rela-
tionships affecting who can and cannot access
resources. Indeed, WUAs can act as vehicles for
elite capture of these resources.
This report therefore supports the conclu-
sion of IEG’s recent sector study of irriga-
tion (IEG 2006b), which states that the Bank
has been overly optimistic regarding what
may be achieved through WUAs and has un-
derestimated the need for continued gov-
ernment involvement. This finding applies to
technical issues such as the design of rehabilita-
tion works as well as to the ability to intervene in
water disputes and—for the time being at least—
to continue subsidization of system operation.
WUAs are still establishing their roles, so they
may yet fill their potential, but support will be re-
quired for this to succeed.
Lessons Should the Bank avoid financing large irrigation
schemes? This has been the trend, but to abandon
an activity with a large poverty impact is to side-
step the issues involved rather than to resolve
them. For areas where there is unexploited irri-
gation potential, there are still lessons to be learned
about the need to undertake realistic appraisals
to avoid financing projects that are not viable.
On selecting irrigation investmentsMany previous lessons “learned” still remain to be
learned:
• Construction delays undermine the economic
viability of investments, so large-scale con-
struction needs to be accompanied by special
measures (such as those currently being
adopted in Andhra Pradesh) to avoid delays and
cost overruns.
• Unrealistic appraisal estimates increase the
likelihood of poor investments.
• The method of risk assessment used in ap-
praisals understates the actual risk that the
project may not be viable.
• All benefits and costs need to be valued.
On PIM• Support for bottom-up development needs a
strong lead from the top.
• PIM may alleviate the problems of large-scale
irrigation schemes, but it cannot eliminate
them.
• WUAs have only limited means to effectively re-
solve water allocation disputes.
• PIM is not intrinsically pro-poor.
On the impact of irrigation• Irrigation raises output through higher yields
and cropping intensity and stabilizes produc-
tion, all of which contribute to lower poverty.
But irrigation investments constitute a size-
able subsidy to farmers who are not among the
poorest or even the poor.
• The direct income benefits to farmers from
obtaining irrigation provide the potential for full
cost recovery of the recurrent costs of the ser-
vices provided.
• Incentive structures, including water pricing
policy, can encourage water wastage.
x v i i i
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Chapter 1Evaluation Highlights• It is paradoxical that irrigation proj-
ects are uneconomical and havenegative social and environmentalconsequence—yet have had strongeffects on poverty reduction.
• Overcoming this paradox requiresaddressing the causes of construc-tion delays and cost overruns andensuring adequate operations andmaintenance.
• Andhra Pradesh did this through leg-islating water user associations, butthe approach faltered and in anycase could not tackle all the prob-lems in the system.
Introduction
This improvement has been made possible by a 30
percent increase in per capita grain production as
irrigated area increased 250 percent. In India, the
expansion of irrigation and adoption of new crops
(the Green Revolution) went hand in hand, with
gross cropped area2 increasing from 23 million to
90 million hectares (World Bank 1997, p. i.).
The Irrigation ParadoxMany studies show the link among irrigation,
higher farm incomes, and substantial multiplier
effects benefiting other rural residents. In India,
poverty rates in irrigated districts are one-third
what they are in districts without irrigation (World
Bank 2005).3 An analysis of poverty trends in 36
villages in Andhra Pradesh found that irrigation
was the main explanatory factor for one-quarter
of all households that had escaped poverty. In
some areas, that figure was more than half
(Krishna and others 2004). Poverty rates are lower,
and social indicators higher, in the coastal re-
gions of the state that have the most irrigation
coverage and have benefited most from major
irrigation schemes.
The Independent Evaluation Group’s (IEG) recent
review of the irrigation subsector identified five
channels through which irrigation can reduce
poverty: direct impact on income, including own
consumption, of poor farmers; increased agri-
cultural employment and wages; multiplier ef-
fects through irrigation-induced growth; lower
food prices; and empowerment through in-
creasing the assets of the poor and their access
to decision making through community organi-
zations (IEG 2006b).
Despite the apparent role that it could play in
achieving the Millennium Development Goals,
irrigation has fallen out of favor. For exam-
ple, World Bank lending for irrigation
peaked in the 1980s and has been
falling since.4 In India, lending fell
from the mid-1980s, and the number
of active projects has remained stag-
nant at around 10 in each year over the
past several decades. Lending rose again after the
mid-1990s, but with a focus on rehabilitation
and institutional reform rather than on new
construction.
This shift in attitude was prompted by, among
other things, inefficiencies within irrigation systems
resulting from poor operations and maintenance
(O&M). Inefficiencies such as cost overruns have
meant that where ex post rates of return have
been calculated, they frequently fall well short of
10 percent (Thakkar 1999).5
In India, for example, IEG rated 8 of 10 irrigation
projects over the years 1991–95 unsatisfactory
on outcome. All 10 suffered implementation
delays and had economic returns well below
the appraisal estimates. The problems included
overly optimistic appraisals with unrealistic time-
frames; poor project design, poor management,
and bad contractors; rent seeking associated with
Irrigation has been a powerful force for poverty reduction and has elimi-
nated the scourge of malnutrition for millions of people. Across the de-
veloping world, average daily calorie availability has risen from 1,930 to 2,6701
over a 42-year period (1961–2003).
3
Though irrigation hasbeen a powerful force forpoverty reduction, it hasfallen from favor.
large-scale public investment programs; and ini-
tiation of follow-up projects before addressing ex-
isting problems.
These problems persisted despite reform attempts
that included the adoption of a more holistic ap-
proach called command area development (CAD).
CAD sought to develop the command area below
the project outlets and organized farmers’ groups
to take part in system management. These
farmer’s groups were the forerunners of the water
user associations (WUAs) that feature in the Bank’s
current support for participatory irrigation man-
agement (PIM), usually with a view to eventual ir-
rigation management transfer (IMT).
Irrigation lending has also been criti-
cized for adverse social and environ-
mental impacts. Some irrigation
projects have been associated with in-
adequate resettlement programs (IEG
1998). Falling groundwater levels and
waterlogging, as well as increasing salinity, have
been among the negative environmental impacts
of irrigation projects, impacts that undermine
the benefits of the schemes themselves. These
problems are acknowledged in the opening para-
graph of the Bank’s 2004 water resources sector
strategy, which states that “50 percent of the
world’s wetlands have disappeared in the past
century . . . with water tables already deep and
dropping every year, and some damaged perma-
nently by salinization” (World Bank 2004a).
It may seem that there is an “irrigation paradox,”
in that an investment that has done so much to
reduce poverty has come to have such a bad
name. And within this paradox is another, which
is that large-scale irrigation schemes (see box
1.1) have attracted the most criticism, even though
they are precisely the schemes supporters say
are needed to bring the benefits of ir-
rigation to all those who need it.
The World Bank attempted to over-
come this paradox by changing its ap-
proach to irrigation. Its new approach
was embodied in its 1993 water resources man-
agement policy, which addressed the negative
environmental impacts and lack of sustainability
that had become apparent. The new strategy
called for an integrated approach to water re-
sources management and for reforms to improve
the relevance, effectiveness, and sustainability of
operations in the main water subsectors.
A 1997 report described these problems in the In-
dian context (World Bank 1997). The emphasis on
new construction had diverted attention from
quality of construction and operations and main-
tenance (O&M). What funds there were for O&M
were inadequate and mostly used for the salaries
of government agents, but government irrigation
departments were inefficient and unaccountable.
Poor-quality irrigation systems set off a vicious
cycle in which cost recovery was weak, as farm-
ers were unwilling to contribute much for poor
services. Further expansion of the system was
constrained physically (many states exhausted
their irrigation potential), environmentally (grow-
ing environmental problems limited productivity),
and fiscally, with irrigation attracting a lower
budget share. The proposed reform program en-
compassed IMT and PIM, privatization of O&M,
raising private funds for new investments, and
appropriate pricing of water and related inputs.
The Bank argued that these far-reaching reforms
could support a second revolution of agricultural
growth through productivity enhancement; the
potential from the resulting scenario was much
greater than from continued area expansion.
IEG’s evaluation of the 1993 strategy found it had
continuing relevance but had been only partially
implemented (IEG 2002). In the irrigation sub-
sector, progress had been made with IMT, but lit-
tle progress was made on cost recovery and
virtually none on addressing other subsidies that
encouraged excessive water consumption,6 such
as low rural electricity tariffs. In line with these find-
ings, the 2004 Water Resources Sector Strategy
(World Bank 2004a) reinforced these messages,
promoting further reform while stressing the im-
portance of water resources for poverty reduction
and emphasizing the need for reforms to en-
courage efficiency in resource use. The strategy
4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Irrigation also haspotentially adverse social
and environmentalimpacts.
Attempts to overcome theparadox have addressed
the reform agenda inwater resourcemanagement.
added that more needed to be done to enhance
the poverty impact of irrigation (World Bank
2004a).
The change in emphasis of Bank lending for ir-
rigation may be the appropriate response to
the “irrigation paradox.” Indeed, there may be
no paradox at all: irrigation achieved great ben-
efits in the past, but now returns are falling as
the scope of major productivity increases has
been exhausted. This would suggest there is
no longer scope for new investments. That is
not entirely true, even in South Asia, and cer-
tainly not in Africa. So consideration of the im-
pact of investing in new infrastructure remains
relevant.
This IEG impact study looks at two Bank-
supported irrigation projects: the Second and
Third Andhra Pradesh Irrigation Projects—AP II
(1988–96) and AP III (1996–2004). The goal of AP
II was to develop a new command area along the
Srisailam Right Bank Canal (SRBC) and to reha-
bilitate and extend the command of the Kakatiya
Canal under the Sriramasagar Project (SRSP).
However, because of various problems, the works
envisaged under AP II were not completed, so AP
III was implemented in the same areas.
I N T R O D U C T I O N
5
Box 1.1: Irrigation: The Basics
Irrigation sources may be either surface water or groundwater.Surface water irrigation draws from natural rivers, artificialcanals, or tanks. Canal irrigation is a treelike system consistingof a main canal and major branches called distributary canals,from which minor (or tertiary) canals spread to reach field canals(or channels).
A canal system is divided into reaches. The head reach is near-est the water source, and the tail reach is farthest from it; the mid-dle reach is between the two. The whole area covered by theirrigation system is called the command area.
Dams are used throughout the system to control flow. Themain canal uses a dam to divert river water from its usual course so it flows into the canal. A gate at the head of eachdistributary and minor canal is used to control the flow of water through the system. Long main canals may have addi-tional dams along their length to control the water flow in thelower reaches. Tanks are “minidams,” collecting water fromvarious sources; that water is then distributed by a minor canalnetwork. Pumping is not usually required using canal or tankwater, as that water flows from the field channels straight intothe fields.
In Andhra Pradesh, irrigation schemes are designated asmajor, medium, and minor and according to their command area.Major schemes are those larger than 10,000 hectares (ha), mediumschemes are between 2,000 and 10,000 ha, and minor schemesare smaller than 2,000 ha.
Groundwater irrigation uses boreholes or tubewells to extractwater from aquifers using pumps. Pumps are now usually elec-tric or diesel operated.
The benefit of large-scale irrigation schemes is that they dis-tribute water over a large area, so that individual areas are lessreliant on local water availability. Rain-fed agriculture that doesnot use irrigation is most sensitive to variations in water supplybecause it depends on rainfall fluctuations. But locally based ir-rigation also depends on the local watershed, and wells may dryup if the groundwater level falls too low.
Irrigation involves complicated resource distribution issues.Capturing water for irrigation deprives users who have been ac-customed to a more plentiful supply. For this reason, the con-struction of major irrigation schemes has become a source ofdispute between states. Yet the presence of a canal or tank raisesthe groundwater level in surrounding areas through seepage. In-deed, in Andhra Pradesh some of the older tanks have been con-verted into percolation dams by blocking the outlets. (A percolationdam or tank is a structure to collect water that then filters into thesoil, recharging the aquifer.)
The costs of irrigation are the cost of constructing the waterstructure, maintenance costs, and charges for water use. In Indiamost of these costs have traditionally been borne by the govern-ment. Revenue is collected from users by the water cess, whichis charged according to the area cultivated rather than the amountof water used.
Amount(US$ ICR/PCR and
Project Dates millions) Command PPAR ratinga
Pochampad Irrigation 1971–79 41 Pochampad Dam and command, now called SRSP n.a.
Godvari Barrages 1976–81 45 Barrage to replace weir for Godvari Estuary n.a.
AP Irrigation and CADComposite Project (AP I) 1976–85 136 Nagarjunasava Reservoir n.a.
AP II 1988–94 118 SRBC and SRSP ICR: Highly unsatisfactory
PPAR: Highlyunsatisfactory
AP III 1997–2004 272 SRBC and SRSP ICR: Satisfactory
Economic Restructuring 1999–2006 142b Support for reform and institutional development Project not yet closedProject across the stateSource: IEG project database.Note: AP = Andhra Pradesh; PCR = Project Completion Report; ICR = Implementation Completion Report; PPAR = Project Performance Assessment Report; n.a. = not applicable. a. The Bank’s ICR, formerly known as the PCR and the PPAR, is an independent review prepared by IEG. Ratings were not assigned in earlier projects. b. Irrigation component only.
The Andhra Pradesh ContextAndhra Pradesh has an extensive irrigation net-
work, including infrastructure constructed with
World Bank support (table 1.1), which suffered
many of the problems outlined above. By the
mid-1990s the state’s irrigation sector was clearly
in crisis. The actual area irrigated had begun to fall
during the 1980s as lack of maintenance left many
distributaries and minor canals unusable. By 1995,
1.54 million ha were irrigated by canals, less than
25 years earlier and more than 15 per-
cent less than the peak reached in 1983.
Under both major and minor schemes
(see box 1.2 for definitions), the state
has an irrigation gap—that is, a gap between the
area for potential irrigation with the existing sys-
tem and the area actually irrigated—of about 1.5
million ha.
The government of Andhra Pradesh produced a
white paper in 1996 identifying a series of prob-
lems, including a decline in net irrigated area as a
result of poor O&M, inefficient and inequitable
water distribution, low yields, and low farm income.
These problems were to be addressed
by the Andhra Pradesh Farmers’ Man-
agement of Irrigation Systems Act
(APFMIS Act), passed by the Legisla-
tive Assembly in April 1997. Water
charges were more than tripled in the same month7
and more than 10,000 WUAs created. The posts in
the WUAs were filled through statewide elections
on a single day in June 1997. Only five months later,
174 distributary committees were created.8 With
the APFMIS Act and the subsequent creation of the
WUAs, Andhra Pradesh was one of the leading
areas in the world implementing PIM.
An early Bank study showcased the experience
(Oblitas and others 1999). The creation of so
many WUAs and the staging of statewide elec-
tions was indeed quite an achievement. How-
ever, nearly half of the elections were uncontested,
at least in part because the government offered
an incentive of 15,000 rupees (Rs 15,000) to each
WUA that unanimously elected all its committee
members and president (Mollinga, Doraiswamy,
and Engbersen 2001).
The philosophy of PIM was consistent with the
Janmabhoomi program launched by the ruling
Telegu Desam Party in 1997. Janmabhoomi was
labeled by the state government as a “people-
centered participatory development process”
through which the chief minister wanted to trans-
form the role of the government from service
provider and regulator to facilitator. Under such
a philosophy, the role of the Irrigation and Com-
6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Irrigation in AndhraPradesh was in crisis by
the mid-1990s.
Sustainability problemsin Andhra Pradesh were
addressed throughlegislation creating WUAs.
Table 1.1: Bank-Supported Irrigation Projects in Andhra Pradesh
mand Area Development Department (I&CADD)
would shift from management of the irrigation sys-
tem to provision of technical support to WUAs.
WUAs would now collect water charges, undertake
O&M, and be responsible for management of the
system—up to command level—through the
three-tier structure of WUAs, distributary com-
mittees, and project committees. However, many
observers saw a political motivation behind the
creation of new structures at the grassroots level;9
they provided an opportunity to build a power
base for the Telegu Desam Party outside of the for-
mal structure of village governments (the Gram
Panchayats), which were dominated by the Con-
gress Party.
However, changing political circumstances tem-
porarily reduced the momentum of PIM. New
WUA elections were not held on schedule when
the terms of office of the WUA officers elected in
1997 expired; many WUAs became moribund
until elections eventually took place in 2005–06.10
This momentum has been somewhat restored
in the last two years. However, the new Congress
Party government came to power on a platform
of promoting agricultural development, central to
which are new irrigation schemes. Unlike many
other areas in India, Andhra Pradesh has reached
only about half its irrigation potential. The latest
five-year plan envisages substantial investments in
new irrigation (appendix table D.1). But the Bank
has been reluctant to commit to a substantial
program without improved efficiency in the ex-
isting system. This evaluation thus feeds directly
into the very relevant question of the viability of
large-scale irrigation investments in Andhra
Pradesh. Central to the study, therefore, are the
underlying questions on the role of the Bank in
promoting PIM and how the adoption of PIM im-
proves the viability of irrigation investments.
Evaluation QuestionsThe study considers six questions:
• Do investments in irrigated agriculture increase
production in Andhra Pradesh?
• What are the impacts of PIM on incomes and
poverty alleviation, and who benefits?
• Is investment in major canal irrigation
economical?
• Was the Bank’s support for canal irrigation rel-
evant and appropriate?
• Do WUAs facilitate greater participation by the
poor and by disadvantaged groups in water
management and access to irrigation?
• Are WUAs an effective means of providing sus-
tainable O&M?
Study Approach and OverviewThe study uses a quasi-experimental design, as did
IEG’s recent impact evaluations (IEG 2006b, 2005,
2004, 2000). Undertaking an impact study for a
project that had closed and for which there were
no baseline data was a challenge.11 But although
the SRBC project closed in 2004, many farmers
were still not connected to water in
the new command area. The IEG team
was informed that approximately
44,000 ha of the SRBC command would
be irrigated by the 2005 kharif season,
with another 21,000 ha to be connected
I N T R O D U C T I O N
7
Box 1.2: Some Local Terminology
Caste: Castes are hereditary social groups, the members of which weretraditionally restricted in the activities they could perform. The 1950 con-stitution of a newly independent India included a schedule of castes andtribes thought to be the most disadvantaged—hence the name scheduledtribes and castes—who benefited from reservation of certain positions,including members of Parliament, where they alone were allowed to stand.The 1980 Mandal Commission proposed to extend reservation to other back-ward castes, that is, group less well off than the remaining other castes,though not as underprivileged as the scheduled tribes and castes. Theserecommendations were implemented, against considerable protests, in 1990.
Mandals: Each state is divided into districts and each district further di-vided into subdistricts, which in Andhra Pradesh are called mandals.Andhra Pradesh has 18 districts and 786 mandals.
Rain and Cropping Season: There are two main cropping seasons inAndhra Pradesh: kharif, for which planting is immediately after the mon-soon in July/August, and rabi, in the first months of the year. Rain-fed areashave only a kharif season, whereas those with irrigation may plant inrabi and possibly a third, summer, season. Alternatively, some crops canbe cultivated twice a year but do not fall within a specific season.
Andhra Pradesh still hasirrigation potential, butthe Bank has drawnattention to the need toaddress efficiency.
the following year. This phasing-in of irrigation
suited IEG’s sample design, which was based on
the “pipeline approach” of selecting control areas
to be treated that were as yet untreated (see
appendix A).12
The first round of the survey, carried out
in June 2005 (see box 1.3), would in ef-
fect be a baseline for areas to be irri-
gated (treated) for the first time in the
month immediately after the survey.
The second round of the survey, in
March 2006, would capture the impact of the first
two cropping seasons (kharif and rabi) for these
newly irrigated areas. Unfortunately for both the
study and the farmers, the Irrigation Department’s
figures were something of an exaggeration (see
chapter 2). A far smaller area of the SRBC command
had been irrigated—and this by test waters—and
the full command was only due to be completed
by mid-2007. The lack of irrigation under the SRBC
command made it unsuitable for the survey; in-
deed, it means that impact estimates for the SRBC
command cannot be based on actual data.
The survey was carried out in the Warangal dis-
trict, toward the tail reaches of the SRSP command
below Lower Manair Dam.13 This system had been
both rehabilitated and extended under AP II and
AP III. But five years of drought meant that in the
2004–05 season no water was released in the
SRSP canal for the first time since it had opened.
Groundwater levels were low so that other irri-
gation sources were also yielding less than in pre-
vious years.
In contrast, rainfall from the monsoon season of
2005–06 was very good. The survey provides
ample evidence of how the rains improved agri-
cultural performance and reduced poverty. But
these factors do make it more of a challenge to
identify the impact of the irrigation projects. This
difficulty is compounded by the lack of mea-
surement of water releases below the head of
the distributary canal (and even these data are not
compiled centrally), so there are no accurate
means of measuring water use. For evaluators, the
disruption of canal water supplies in 2004–05
provides a “natural experiment,” 14 whereby a
before-versus-after comparison (in fact, after ver-
sus before) of normally irrigated areas can provide
additional indicators of irrigation impact.
The use of data from two survey rounds allows for
double-difference estimates of project impact to
be made, that is, impact calculated as the change
in outcome in the project area less the change in
outcome in the comparison group (see appendix
C). Many households received canal water through
recharging of the village tanks. Hence, project
beneficiaries are identified as those using both
canal and tank waters.
Comparisons between beneficiaries and non-
beneficiaries were made at three levels: plot,
household, and village. For the plot and house-
hold comparisons, comparisons were made be-
tween those who had newly acquired canal or tank
irrigation between the two rounds of the survey
and those who had not. Those who had not made
up two separate comparison groups: those who
had no irrigation at all in 2005 or 2006 and those
who continued to use the same source. The for-
mer is the more appropriate control and is best
compared with the subsample of the treatment
group that had no irrigation in 2005. However, the
sample size does not always permit this compar-
ison. The comparisons of village-level averages
compare villages that received canal irrigation for
8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Box 1.3: The IEG Survey
The IEG survey consisted of two rounds: one in June 2005 and the sec-ond in March 2006. The first round, just before the 2005 kharif season started,asked questions about production in three preceding seasons (2004 kharifand 2004–05 rabi and summer 2005). The second survey asked about thethree subsequent seasons: kharif in 2005, rabi in 2005–06, and expecta-tions regarding summer 2006. Fifty households were surveyed in 18 villages,making a total sample size of 900 households. The same households wereinterviewed in both rounds. A village questionnaire and a questionnairefor the WUA chairman were included. A separate qualitative study teamwas fielded in six villages in March 2006.
Data were collected on income, farm budgets, irrigation, and variousaspects of social and political life in the village.
The analysis presented in this report uses data from a number ofsources—both primary and secondary—in addition to the IEG survey.
The Andhra Pradeshprojects offered an
opportunity for a before-and-after impact
analysis.
the first time between the two surveys with villages
that have no access to canal irrigation.15
The sample for IEG’s survey was selected to fa-
cilitate examination of the impact of new irri-
gation. But the surveyed area is clearly not
representative of the project area as a whole; the
area below Lower Manair Dam is notably different
in cropping patterns because of a relative contin-
ued water scarcity. Therefore, the impact analysis
in this report is based on a number of different
sources, such as the AP III baseline data, various
data from the government of Andhra Pradesh,
and the figures used by World Bank operational
staff in the completion report for AP III.
Overview of the ReportChapter 2 describes the two projects studied: AP
II and AP III. The following two chapters discuss
the workings of WUAs (chapter 3) and the eco-
nomic impact of the project (chapter 4). Chapter
5 lists lessons learned.
I N T R O D U C T I O N
9
Chapter 2Evaluation Highlights• AP II suffered from a faulty and in-
complete appraisal and failed.• AP III aimed to complete the work
started by AP II.• AP III adopted a participatory pro-
cess that relied on WUAs for irriga-tion management.
• The works supported by the projectexpanded the area under irrigation,but many farmers have yet to receivewater, nearly 20 years after construc-tion started.
• Despite improved service, water al-location issues still exist; for example,irrigation water often does not reachtail-enders.
The Projects
Second Andhra Pradesh Irrigation ProjectFirst, SRSP was a renovation and extension of the
system, with modernization of 165,000 hectares
(ha) in the command area already supported by
the Bank under the Pochampad Irrigation Project
(1971– 79). A further 163,000 ha were to be de-
veloped below the Lower Manair Dam, located at
kilometer (km) 146 on the Katakiya canal. Second,
the SBRC irrigation scheme involved construct-
ing a main canal 143 km long with a capacity of
750 cubic feet per second, with block develop-
ment to irrigate an area of about 65,000 ha.
These main canal works were to be supplemented
by supporting activities, such as the provision of
training facilities and training of professional and
support staff required for project implementa-
tion, training of farmers, resettlement activities,
and construction of feeder roads.
In accordance with the project objectives, the
principal project benefits expected were higher
agricultural value and farm employment. However,
these benefits were not realized because the proj-
ect ran into serious construction problems. There
were both substantial cost overruns (of 326 per-
cent for SRSP and 175 percent for SRBC) and de-
lays in carrying out work. Although some major
construction was implemented, or at least partly
implemented, no contiguous portion of new or
rehabilitated canal or command area was com-
pleted before the project closed. Just over one-
third (38 percent) of the planned civil works were
completed in the SRBC command, and only about
one-fifth (21 percent) of those planned in SRSP
were completed. Thus, no irrigation—either mod-
ernized or new—was provided under AP II.
The project’s failure can be attributed
to problems in both design and im-
plementation. The design underesti-
mated costs, notably for the amount of
hard rock excavation that was required to con-
struct the SRBC main canal. The problem was
made worse by government-estimated unit costs
(used in the project cost estimates) that were
well below the market values for such works. The
selection of contractors was poorly done, and
they performed badly, exacerbating the cost over-
runs. Project documents also speak of “rent seek-
ing” associated with construction activities,
although no direct evidence is provided.1
Project design also took inadequate account of two
issues. The first was the difficulty of acquiring
land. Second, because water supply was limited,
the state government’s policy was that all new sur-
face irrigation projects in the two regions would
be designed and operated to satisfy the require-
ments of irrigated dry crops, requiring an en-
forced change in the cropping pattern because of
reduced water allocation (a policy known as lo-
calization, which dated from colonial times).
Specifically, under AP II, water would be required
to fill Lower Manair Dam to provide irrigation in
the newly developed areas below the dam. This
water previously had been available to farmers in
AP II started in 1988 and was the fourth Bank-supported irrigation proj-
ect in Andhra Pradesh (see table 1.1). With the objective of increas-
ing agricultural production and farm incomes, AP II was to bring
irrigation to two poor, drought-prone areas through two subcomponents.
1 3
AP II ran into severeconstruction problemsand failed.
the SRSP command above Lower Manair Dam, so
those farmers were accustomed to irrigated wet
agriculture, notably paddy, which is the preferred
crop in the region.
However, attempts to force changes
in cropping patterns have a poor record
in Andhra Pradesh, including the fail-
ure of this policy in an area adjacent to
SRBC in the first irrigation scheme in
Andhra Pradesh. These difficulties were
a manifestation of a problem well en-
trenched in Indian irrigation. Schemes
are designed with the widest possible coverage,
partly to help ensure food security and partly for
political reasons. But the system design is overly
optimistic, not taking account of likely cropping
patterns and actual system losses. Hence, there
is insufficient water for all users. In the absence
of allocation through pricing, the result is head-
enders taking more water and tail-enders going
without.2 There has been conflict at times.
Overall, the Bank’s preproject assessment of AP
II was severely flawed: the Bank accepted the
government’s inaccurate cost estimates and a de-
sign the Bank itself later questioned. Moreover,
the Bank also accepted projections of the area to
be covered, future yields, and crop mix, all of
which proved optimistic—a situation that should
have been expected given earlier experience.3
Substantial problems were also caused by the
Bank’s management of the project. Design
changes were required after the first contracts had
been issued. It took time to come up with the re-
vised designs, but the Bank then commissioned
studies that supported the original design. There
was a protracted dispute over the sufficiency of
the water allocation for SRBC, during
which the Bank informally suspended
the credit. Finally, lack of familiarity
with Bank procedures among govern-
ment officials slowed procurement.
With slow progress, the Bank cancelled $145 mil-
lion of its $271 million contribution to the project.
However, there was agreement that a new project
(AP III) would support completion of the works.
Third Andhra Pradesh Irrigation ProjectThe overall objective of AP III (1996–2004) was
to complete the irrigation development and re-
habilitation work begun under the previous proj-
ect and thus realize the potential for increasing
agricultural productivity and rural incomes in
two economically backward regions of Andhra
Pradesh. This was to be achieved through a num-
ber of activities.
The project was scaled back, with the following
elements dropped: irrigation development and re-
habilitation works (80 percent of cost) for com-
pletion of the irrigation network and feeder roads
of the 65,000-ha SRBC irrigation project; rehabil-
itation of 253,000 ha of the SRSP scheme’s irri-
gation system, of which 88,000 ha were below
Lower Manair Dam, up to km 234; and extension
of the canal beyond km 234.
Agricultural support services (2.5 percent of cost)
would foster agricultural diversification and pro-
ductivity through irrigation agronomy research
and improved command area extension services,
and improve irrigation services by PIM through
the establishment of WUAs. This component
had three subcomponents: (i) an irrigation agron-
omy program, (ii) a WUA promotion program, and
(iii) a farmer training program. Support to WUAs
in the SRSP command was, for the Bank, piloting
an approach that was to be expanded statewide
in the irrigation component of the Andhra Pradesh
Economic Restructuring Project.4 Activities sup-
ported under AP III were expected to serve as a
model for the rest of the state with respect to as-
pects such as collection of water charges.
The project also included a resettlement and re-
habilitation program (5 percent of cost), an en-
vironmental management plan (7.5 percent of
cost), and dam rehabilitation and safety assurance
works (5 percent of cost).
Project Implementation
SRSPThe rehabilitation works for SRSP were com-
pleted. This resulted in an increase in the carry-
ing capacity of the main canal and the command
1 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Costs wereunderestimated and
the difficulty ofacquiring land and
of achieving a changein cropping patterns
was neglected.
AP III was undertakento complete the
unfinished work of AP II.
irrigated area of 115,000 ha from 1995–99 (fig-
ure 2.1); by 2005, the area irrigated above Lower
Manair Dam had reached the target of 165,000 ha.
Rehabilitation works above Lower Manair Dam
have enabled water to flow for a greater distance
down distributaries and minor canals, so that
more areas get water. Actual irrigation in the SRSP
command as a percentage of that planned rose
from an average of 50 percent in 1995–96 to
nearly 90 percent five years later (appendix table
D.2). The works below Lower Manair Dam have
been completed, with the most recent data show-
ing an irrigated area of 90,000 ha, slightly above
the 88,000 ha expected at appraisal.5
However, it is likely that the additional irrigated
area is in fact short of that expected under AP III.
The figures reported here are for the area I&CADD
believes will be irrigated, which overestimates
the actual area irrigated. Revenue Department
estimates of irrigated area are typically 40 percent
lower than those of I&CADD (Jairath 2000).6 Al-
though Revenue Department estimates may be bi-
ased downward,7 I&CADD figures are likely to
overestimate the area by at least 20 percent. The
actual area covered is invariably less than the po-
tential reported, owing to the failure of water to
reach tail-enders because of system losses (a well-
known problem, confirmed by the field work for
this study) and excess use by farmers in higher
reaches.8
Excess use in higher reaches is discussed later in
this report. Regarding system losses, the “baseline”
survey found the water flow in the main canal to
be 90 percent of capacity on average, ranging
from 100 percent in higher reaches to 76 per-
cent toward the tail end. The average capacity of
the distributaries is less. The failure to realize full
capacity results from buildup of silt, land slip-
pages around cuttings and embankments, and
damaged linings. These data were collected shortly
after the completion of the rehabilitation financed
by AP III, so the situation is likely to have wors-
ened since then.
Nonetheless, the greater carrying capacity of the
main canal has made possible the expansion of the
command area; under SRSP Stage II,
the state government is extending the
main canal a further 62 km, the dis-
tributaries of which will eventually ir-
rigate 440,000 ha. In the areas covered
by IEG’s survey, which included the newly irrigated
areas between km 234 and km 284 of the main
canal, the area irrigated by canal or tank doubled
from 13 percent of all cultivated land in 2004–05
to 26 percent in 2005–06 (figure 2.2; see also ap-
pendix table D.3).
The bulk of this increase is from tanks. This is
partly because tanks are quickly recharged by
good rains, such as occurred in the monsoon be-
tween the two survey rounds. But it is also because
in the majority of the surveyed villages, the canal
water is fed into the tank and distributed via the
tank’s minor irrigation system. This practice is
very evident in those villages newly connected to
the canal system between the two rounds: the per-
centage of cultivated area receiving
water from a tank jumped from 9 to 53
percent, compared with 0 to 8 percent
receiving water directly from the canal.
South India has a well-developed
T H E P R O J E C T S
1 5
The rehabilitated worksfor SRSP successfullyincreased the irrigatedarea.
The area covered may stillbe below what wasexpected because offlawed estimates.
Figure 2.1: Irrigated Area in SRSP (000s of ha)
0
50
100
150
200
250
300
1995–96 1996–97 1997–98
Are
a
1998–99 1999–2000
KharifRabi
Source: SRSP baseline.
Notes: No water was released in rabi 1997–98. ha = hectares; SRSP = Sriramasager Project.
system of tank irrigation, with some tanks dating
back centuries; existing tanks are sometimes in-
corporated into the canal system. However, as dis-
cussed below, the storage of canal water in tanks
has implications for water allocation and thus can
be contentious.
SRBCIn SRBC, although the main canals and distributaries
were completed, the minor works were only about
20 percent complete as of June 2005, one year
after the project closed. The state government is
completing the works with its own funds, esti-
mating that the works will be finished by 2007. In
2005, some test water had been released, and farm-
ers had breached the main canal to ir-
rigate their fields. Officially, only 7,880
ha had been irrigated, compared with
the planned amount of 65,000 ha.
As noted in chapter 1, the IEG study originally in-
tended to undertake a survey in the SRBC com-
mand based on information from I&CADD (also
given in the Bank’s Implementation Completion
Report), that 44,000 ha had already been irri-
gated. This study found that the 44,000 ha refers
to irrigation from all sources. Counting all irriga-
tion in the command area, whether sourced by the
canal or not, is apparently common practice
(Jairath 2000) and contributes to the overesti-
mation of irrigated area and consequent overes-
timation of benefits and underestimation of the
true per hectare costs of irrigation through major
irrigation schemes.9
These delays seriously affect the project’s viabil-
ity, even if it eventually reaches its full potential.
The Bank’s Staff Appraisal Report (SAR) estimated
the economic rate of return (ERR) for the SRBC
subcomponent to be 13.7 percent. Assuming the
benefits are as stated in the SAR but subject to
the delays that have in fact taken place, this return
drops to 4.1 percent and becomes negative (–1.0
percent) once the original investment costs under
AP II are taken into account.10
SummaryThe AP II project was to newly irrigate 65,000 ha
under SRBC and 163,000 ha under SRSP, as well as
to modernize the system serving a further 165,000
ha under SRSP. However, problems in implemen-
tation meant that no works were finished, so no
farmers received irrigation as a result of the proj-
ect. With positive costs and zero benefits, the rate
of return to the project is negative infinity.
AP III was to complete the works started by AP
II. At the time of this study, the bulk of the con-
1 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Delays in the completionof SRBC have undermined
project viability.
Figure 2.2: Irrigation Sources, 2005 and 2006
0
10
20
30
40
50
60
Canal 2005NoneTubewellBoreholeTankCanal
2005 2006
Canal 2006 Tank 2005 Tank 2006 Tube 2005 Tube 2006
Perc
ent o
f cul
tivat
ed a
rea
Source Source
0
10
20
30
40
50
60
Perc
ent o
f cul
tivat
ed a
rea
None New Existing
Source: Appendix table D.2.
nections in the SRBC command had yet to be
made. They are currently expected to be com-
pleted by June 2007, nearly 20 years after the
work started.11 Such serious delays, which per-
vade irrigation investments across India and else-
where,12 seriously undermine the viability of the
investment in the SRBC command, which will
be negative even if all intended benefits are even-
tually realized.
Progress has been made in AP III, where the in-
creased carrying capacity of the main canal has
made possible a further extension of the system.
However, the areas irrigated have not reached
those planned at appraisal, and, even immedi-
ately after rehabilitation, actual water flow in the
main canal was 90 percent of capacity (and less
in the distributaries) as a result of siltation, land
slippages, and damaged linings.
T H E P R O J E C T S
1 7
Chapter 3Evaluation Highlights• PIM is classified as pro-poor be-
cause it involves the poor in irrigationmanagement and raises agriculturalincomes and employment.
• Both membership and leadership ofWUAs are biased toward those whoare better off.
• The design of the irrigation systemand management by the WUAs havein some cases adversely affectedwater distribution.
• Only a small percentage of non-WUAmembers believe water is fairlydistributed.
• The reforms have not ended the cycleof poor irrigation services and un-willingness to pay for O&M.
Building a canal for irrigation in India. Photo by Ray Witlin, courtesy of World Bank Photo Library.
2 1
Water User Associations
Origins of PIM in Andhra PradeshA first attempt to change this in India came in 1973
with the initiation of CAD, which relied on farm-
ers’ organizations for water management and
O&M below the distributary level. During the
1970s and 1980s, the Indian government pro-
moted CAD across India, with Bank support,
through the National Water Management Pro-
gram. As today, Andhra Pradesh was a leading
state in those reforms, with the former commis-
sioner for SRSP successfully lobbying for the
creation of a separate CAD Department and be-
coming its secretary for its first seven years (Siva-
mohan and Scott 2005).
These early experiences offer two lessons. First,
in Andhra Pradesh, more than 3,000 farmers’ or-
ganizations were formed by 1981; they soon dis-
banded, because of insufficient funding and an
inability to solve water distribution issues. The
same problems affect WUAs today.
Second, most reforms are based on earlier policy
changes, which often lay the foundation for later,
more far-reaching, changes. Indeed, despite the
failure of the farmer group aspect of CAD, lobbying
for the approach continued, leading to the pro-
motion of WUAs since 1997.
The Current Wave of PIMThe World Bank supported the promotion of
WUAs through AP III and then across the state
through the Andhra Pradesh Economic Restruc-
turing Project (APERP), which included training
in PIM for WUA and government officials. More
than 10,000 WUAs were created following the
passage of the APFMIS Act, which devolved pow-
ers for system management from I&CADD to
farmers.
Specifically, farmers were to assist in collecting
water fees and would receive 90 percent of the
funds collected to undertake O&M.2 WUAs were
also to determine local water distribution through
the warabandi statement (a schedule of water re-
leases). In principle, farmers were to control water
releases throughout the system through the three-
part committee structure (WUA, distributary com-
mittee, and project committee), but this has not
happened.
The reforms were not without oppo-
nents. Although there was strong sup-
port from the top of both the state
government and I&CADD, lower levels
of the bureaucracy were opposed—one reason
project committees have not yet been created.
Under the changes, the lower-level officials known
as lascars (“gate keepers”), each of whom is re-
sponsible for canal operation and water man-
agement across an area of approximately 800 ha,
were to become accountable to the WUAs. This
proposal was challenged in the courts by 3,500 las-
cars across the state (Raju 2001). Although efforts
were made to build cross-party support, the op-
position accused the government of deferring to
User management of irrigation is not a new idea. Indeed, users tradi-
tionally have managed their own water resources through indige-
nous institutions.1 However, the growth of large-scale schemes financed
by the government shifted management responsibilities to the irrigation
bureaucracy.
A 1997 act devolvedirrigation systemmanagement to farmers.
donor agencies by accepting the conditions of the
World Bank, which included the promotion of
PIM (Sivamohan and Scott 2005).3
This move was expected to increase system effi-
ciency. Many supporters of PIM also claim that it
makes irrigation more equitable, ensuring a
greater spread of benefits to the poor—though
that was not an explicit objective in ei-
ther of the two projects under review.4
PIM is claimed to be pro-poor for two
reasons. First, the involvement of the
poor in irrigation management through
WUAs ensures that irrigation benefits are not cap-
tured solely by those who are better off. Second,
the improved functioning of the irrigation system
under PIM results in higher agricultural income
and employment. This chapter reviews the state
of WUAs nearly a decade after their creation by the
APFMIS Act.
WUA MembershipAccording to the Act, all water users are members
of the WUA; “water users” are defined as those
using water from a government or Andhra Pradesh
Water Corporation source.5 The Act states that cul-
tivating households using such water for any pur-
pose, not just agriculture, are to be members.
However, when asked in June 2005 if they or any-
one in the household was a WUA member, just 10
percent of respondents in cultivating households
replied positively.
This low rate is explained by two factors. First, al-
though virtually all plots are irrigated in one way
or another,6 a minority use canals and tanks for ir-
rigation. Membership in WUAs is notably higher
among the latter: in 2006, half of such households
were in WUAs, compared with less than 10 percent
of households using other forms of irrigation (fig-
ure 3.1). Second, many WUAs had ceased to func-
tion because political problems had delayed new
elections7 and because the lack of water meant that
whether the WUA functioned may be thought ir-
relevant.8 But new elections were held in some vil-
lages between the survey rounds, and water was
released along the length of the SRSP command
in July–August 2005. By March 2006 the propor-
tion of cultivating households claiming WUA mem-
bership had risen to 19 percent.
The second factor behind low participation rates
is that, although all households using the water
resources are meant to be members, households
using canal or tank irrigation are most likely to join
a WUA. The bulk of the increase in membership
has come from tank WUAs, reflecting the fact that
tank irrigation has been revitalized as heavy rain
and canal water have recharged tanks in the sur-
vey area.
The fact that not all farm households participate
in WUAs raises questions about the pattern of
participation. Claims that WUAs ensure the par-
ticipation of the poor in irrigation management
must first carry the caveat that the landless, who
make up the bulk of the poorest, are excluded.
This practice contrasts with the approach adopted
in the Andhra Pradesh Rural Livelihoods Project,
supported by the U.K. Department for Interna-
tional Development. In the Rural Livelihoods Proj-
ect, receipt of funds for watershed development
activities is conditional on first implementing
schemes to benefit the landless and marginal
farmers. Moreover, in principle, these latter groups
2 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Only 10 percent ofcultivating households
were members of WUAs.
Figure 3.1: Households Accessing Canal and TankIrrigation Systems Most Likely to Be in a WUA
0
10
20
30
40
50
60
70
Canal Tank Other irrigation
Prop
ortio
n of
hou
seho
lds
in W
UA
Source of irrigation
2005
2006
Source: IEG survey.
are meant to lead the watershed planning process,
though this rarely happens in practice.
Alternatively, as has happened in other countries,
irrigation projects may include subprojects, such
as fish ponds, that allow the landless to indirectly
benefit from irrigation. Finally, tradable water
rights can be allocated to all households in the
catchment area—strategies to develop this ap-
proach are part of the Bank’s most recent guide-
lines for lending to the irrigation sector in India.
But the bias against the poorest is only partly a re-
sult of the exclusion of the landless. Figure 3.2
shows the proportion of each wealth quartile that
holds membership in a WUA.9 Less than 1 percent
of households in the poorest quartile hold WUA
membership, compared with 28 percent of the top
quartile. Restricting the sample to households in-
volved in cultivation reduces the bias, but not by
much: only 1 in 50 poor households engaged in
cultivation is in a WUA, compared with more than
1 in 3 of the better-off households.
Despite the apparent neutrality of the Act with
respect to caste, landholding, and so forth, there
are also biases in the pattern with respect to
other classifications (appendix table D.4). These
biases are more marked if all households are
considered rather than just farm households,
but they are still present among the latter. Those
households with no education are least likely to
be in a WUA. Large landowners are most likely to
be members, though marginal farmers are more
likely to join than small farmers (figure 3.3).
Households from the top castes are most likely
to be in WUAs and those from scheduled tribes
least likely.10 Using multivariate analysis, the most
robust determinants of WUA membership are
wealth and education (appendix table B.1).
Two factors explain the biases noted. Ownership
of land with access to canal and tank irrigation
is more heavily concentrated among the better off
than is overall land ownership (figure 3.4).11 In
2006, the Gini coefficient for overall land owner-
ship was 0.34, but it was 0.45 for ownership of land
with access to canal or tank irrigation (appendix
tables D.4 and D.5). The second reason behind
the bias is that WUAs operate within the
existing social context, influencing par-
ticipation in ways that often benefit
those who are better off. Such find-
ings are hardly surprising because they
are a well-documented aspect of rural develop-
ment programs in India. The surprise is that more
WAT E R U S E R A S S O C I AT I O N S
2 3
WUAs exclude the poorestfrom decision makingbecause most of them arelandless.
Figure 3.3: Large Landowners Most Likely to BeWUA Members; Marginal Farmers also Active
0Landless Marginal
Amount of land owned
Prop
ortio
n of
hou
seho
lds
with
WU
A m
embe
r
Small Large
5
10
15
20
25
30
Source: IEG survey.
Figure 3.2: Likelihood of WUA MembershipIncreases Sharply with Wealth
01 2 3 4
5
10
15
20
25
30
35
40
All households
Prop
ortio
n in
WU
A
Wealth quartile
Cultivating households only
Source: IEG survey.
was not done to address these known
biases. A better understanding of the
social context in which WUAs operate
would help inform policy so that the
objectives of reform are achieved.
What Do WUAs Do and Who Does It?The APFMIS Act gives WUAs wide-ranging powers,
as they are intended to manage the irrigation sys-
tems.12 However, large parts of these powers have
not been realized, with the Irrigation Department
retaining functions intended to be devolved to the
WUAs. For example, the WUAs are not involved
in revenue collection. Furthermore, funds allo-
cated to WUAs, financed by APERP, were used on
renovation works, with little or none being used
on group development activities.
The Irrigation Department retains control of sanc-
tioning works supposedly under the control of
WUAs; the availability of rents in the contracting
process (Wade 1988), as well as bureaucratic in-
ertia, limit chances that the Irrigation Depart-
ment will give up its own power unless there is a
major political push from the top—an opportu-
nity that was lost once the momentum behind the
reforms faltered. The failure to establish the proj-
ect committees reflects this unwillingness to re-
linquish control.
Water allocationWUAs do have a role in water allocation at the local
level, a process that is frequently marked by rivalry,
both within and between communities. In a ma-
jority of the communities covered by the IEG sur-
vey, most of the irrigation water from the canal is
being channeled into tanks. Tanks have been im-
portant in irrigation in South Asia for centuries,
and incorporating them into the canal system is
a sensible use of existing water distribution chan-
nels. Tanks are also included in the system as bal-
ancing reservoirs (that is, to store water until it is
time for release).
However, evidence from IEG’s field work shows
that storing water in tanks can adversely affect
water distribution. The practice excludes farmers
outside the tank irrigation scheme and some-
times prevents water from reaching those farther
down the canal (see box 3.1): 3 of the 16 vil-
lages surveyed are in the tail reaches of their re-
spective distributaries—none of these villages is
receiving canal water. Another village is not re-
ceiving canal water because the Irrigation De-
partment is opposed to farmers’ requests to direct
the water to the village tank.13 Those farmers not
participating in WUAs—those who are not among
the better off—are most likely those excluded
from benefits, including the half of canal and tank
users who do not participate in WUA activities.
Participation and managementVirtually all WUA members participated in WUA
elections, although this is a minority of those el-
igible to vote, because many eligible households
do not participate in the WUA (table 3.1). One-
third of members regularly take part in meet-
ings;14 somewhat more have participated in O&M.
A smaller number, approximately 17 percent, say
they take part in WUA decision making or be-
lieve they influence water management. This
translates into just 3 percent of cultivators feeling
that they have a say.
Participation in these various activities is not uni-
form across members, but varies according to
the same biases that affect membership. With re-
spect to attending meetings, either at all or reg-
ularly, the multivariate analysis found that the
2 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Figure 3.4: Lorenz Curve for Land Ownership byIrrigation Source
00 10 20 30 40 50 60 70 80 90 100
10
20
30
40
50
60
70
80
90
100
Cum
ulat
ive
shar
e of
land
ow
ners
hip
Cumulative share of households
Canal and tank
Tubewell and borehole
All land
45°
Source: IEG survey.
Note: Households are ranked by a wealth index.
It is not surprising that thebetter off are more involved,
but it is surprising that theprojects did so little to
address the issue.
uneducated are least likely to participate (ap-
pendix table B.2). When it comes to taking part
in decisions or influencing water management, the
wealth variables are significant; indeed, no one in
the bottom quartile replied positively with re-
spect to either of these questions (appendix table
D.7). For voting, as this is near universal, there are
no significant biases among members; again,
members are the minority.
Differential influence is reflected in the pat-
tern of WUA leadership, which is dominated by the
better off. In a survey of 222 WUAs in 22 districts,
88 percent of leaders of canal system WUAs were
from higher castes and 11 percent from backward
castes; just 1 percent were from scheduled tribes
or scheduled castes. Large landowners accounted
for 88 percent of canal WUA presidents.
Elite domination is less marked for tank systems
but still present: 86 percent of presidents are
from higher castes or backward castes and 56
percent are large landowners (Reddy and Reddy
2005). Membership in the WUA committee is
more representative (for example, ORG 2005), but
in many cases the president runs the WUA with
little consultation: in a survey of 214 WUAs, the
majority was not meeting regularly (ORG 2005).15
Irrigation Department officials say that canal design
is determined technically and so cannot be subject
to political interference. Design of the micro-
network ideally is done in consultation
with local communities, but this is not
what happens in practice. Indeed, in one
IEG study village, politically well-connected
families are not among those who will
benefit and are mobilizing people against
the completion of the canal and obstructing
land sales. Nonetheless, there is substan-
tial scope for extracting more water than
envisaged in the system’s design.
The impact of WUA management on water
allocation was captured in a question that
asked if the respondent believed water
to be fairly distributed within the com-
munity. Only 15 percent of non-WUA mem-
bers believe that water is fairly distributed,
compared with 55 percent of mem-
bers, suggesting that WUA members
are able to enjoy better access than
nonmembers. In a multivariate analy-
sis of the response, only one socio-
economic characteristic is significant:
the larger a person’s landholding is,
the more likely he or she is to believe that water
is fairly distributed (figure 3.5; regression results
in appendix table B.3).16
The finding that WUA members are significantly
more likely to say that water is fairly distributed
than nonmembers is robust. In contrast, the pres-
ence of an active WUA in the village (measured by
the village-level WUA membership rate) has at
best no impact on perceived fairness. In some
model specifications it increases the likelihood that
WAT E R U S E R A S S O C I AT I O N S
2 5
Box 3.1: Diversion of Water to TanksLimits the Number of Beneficiaries
In one village in Warangal district, the canal water isdiverted to a tank, depriving two other villages fartherdown the canal of water. This has caused clashes be-tween the villages.
In another village, the canal water flows into fivetanks, each of which has a functioning WUA. There isalso a canal WUA, but it is not operational.
Sources: IEG field visits.
Because much of theirrigation water is storedin tanks managed byWUAs, water distributionhas been adverselyaffected.
Table 3.1: Participation in WUA Activities, 2006 (percent)
All Households with WUAcultivating access to canal membershouseholds or tank irrigation only
Attend meeting 10.8 30.2 58.3
Attend at least six meetingsa year 5.8 15.1 31.0
Participate in O&M 7.3 20.5 39.0
Take part in decision making 3.3 9.1 16.8
Influence water management 3.1 9.4 17.6
Voted for WUA president 17.9 49.0 96.1
Voted for committee chair 17.1 46.5 91.7
Number 557 150 99Source: IEG survey.
people will say that water is not fairly
distributed. The explanation for this
finding is clear. Being in a WUA pro-
vides access to water, which is shared
among WUA members. This process
excludes nonmembers, who thus feel
that water is not being fairly shared.
Hence, the more active the WUA, the
more likely nonmembers are to feel
that water is not shared fairly.
Although it is expected to ensure eq-
uitable water distribution, the WUA
lacks the powers of enforcement needed to do
this.17 Field work found that canal breaches are
common practice, meaning that excess water is
taken off in higher reaches, leaving tail-enders
without enough water. Such matters should be
broached in the WUA and distributary committee
and sometimes are. But the farmers’ groups have
no mechanism for resolving the dispute. Those
breaching the canal may be connected to powerful
families, making it difficult to challenge them in
an open forum. In one surveyed village, the Irri-
gation Department is taking the farmers to court
for breaches.
Higher reaches also use more water than in-
tended as a result of the preponderance of paddy
cultivation (see next chapter). They can also cap-
ture a larger amount of water by diverting the
canal into their village tanks. Of the 18 villages in
the IEG survey, one was not receiving water at all
because of diversion upstream, and two others
were diverting water and were therefore in con-
flict with downstream villages that were not re-
ceiving water.
The distributary committee should resolve these
disputes but cannot.18 Such problems are a man-
ifestation of the fact that decentralization is of-
ten best accompanied by an element of central
control.19
O&MWUAs can claim more success with O&M, al-
though questions might be raised regarding its
sustainability. In SRSP, WUAs were involved in
identifying repair works, though Irrigation De-
partment staff are responsible for repair design.
This approach was adopted statewide in the
late 1990s, partly using World Bank financing.
Most studies are positive about the works carried
out at that time (Jairath 2000; Reddy and Reddy
2005).
It is argued that the involvement of the WUAs in
identifying the works ensured that priority activ-
ities were undertaken and forced a change in
I&CADD’s preferred way of carrying out mainte-
nance on minor canals (Oblitas and others 1999).
One consequence of this was that minor canals
would be in a very poor state before they would
be repaired. Moreover, it is also claimed that cost
savings of about 20 percent were realized as a re-
sult of avoiding rent-seeking by public officials at
various levels (Raju 2001).20
A further cost saving is that the assumption of
these responsibilities by WUAs means that
I&CADD can operate with lower staffing levels.
However, the decisions were usually made by the
WUA president, with little or no consultation with
other committee members, let alone ordinary
WUA members.21 Moreover, WUA presidents are
2 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Only 15 percent of non-WUA members
believe water is fairlydistributed, compared
with 55 percent ofmembers.
WUAs lack the powerthey need to enforce
equitable waterdistribution.
Figure 3.5: Larger Landholders More Likely to ThinkWater Is Fairly Distributed, as Are WUA Members
Like
lihoo
d of
say
ing
wat
er d
istr
ibut
ion
is fa
ir
Land owned (ha)
0.00 5 10 15 20
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
WUA member
Non-WUA member
WUA effect
Source:
themselves frequently contractors: in one study
of eight WUAs, at least six presidents—and pos-
sibly all eight—were contractors who undertook
canal rehabilitation.22
Although it is too early to make a fully informed
judgment, it seems doubtful that WUAs will be ef-
fective in carrying out O&M because the financ-
ing system is unsustainable. Following the reforms,
the government allocated Rs 200 per acre to major
schemes. This figure balances with the water fee
of Rs 200.
But the apparent balance gives a false picture
of financial sustainability for several reasons.
First, 10 percent of the fee revenue goes to local
government rather than any of the water com-
mittees. Second, in major schemes, 20 percent is
allocated to project committees, but there are
no such committees and these funds are not
being paid out.23 Most important, collection re-
mains poor. The amount collected has risen some-
what as a result of tariff increases, but this probably
disguises an actual reduction in collection rates.
The reforms have failed to break the vicious cycle
of poor irrigation services, making farmers unwill-
ing to pay for O&M. Specifically, tail-enders continue
to receive little or no water and are thus unwilling
to pay the water fee. But even farmers in head
reaches frequently do not pay. Collection rates
have been estimated at less than 10 percent of the
amount due (Jairath 2000); data from the SRSP
command show an average of 12 percent for the
period 1998–99 to 2001–02 (I&CADD 2004). Hence,
over the period 1997–2002 only one-third of the
money allocated to canal WUAs and distributary
committees was financed from water charges.24
Moreover, less than half of WUA members believe
that repair works are a WUA responsibility (Reddy
and Reddy 2005). For that reason, prospects are
not good for WUAs sustaining O&M so that the
canal network does not fall victim to the same
problems of poor maintenance. The positive coun-
terpoint is that (as is shown in the next chapter)
in good years, farm income is easily sufficient to
cover even higher water charges.
SummaryThe APFMIS Act was intended to pro-
vide the basis for transferring manage-
ment of the irrigation system from
public servants to farmers. The gov-
ernment at the time was heavily com-
mitted to the reform and provided
support to WUA formation. But reforms
lost momentum with the change of
government, and the transfer has not
taken place. The Irrigation Department
has retained overall control of the command areas
because project committees have not been formed.
WUAs are not involved in revenue collection, and
most farmers see O&M as a government respon-
sibility rather than one for themselves or the
WUAs.
Supporters of PIM claim that it is a channel for rep-
resenting the voice of the poor. This has not been
the case for several reasons. First, the poorest are
excluded because they do not themselves engage
in cultivation. But even among farm households,
a tiny minority of poor households participate in
WUAs, compared with a sizable proportion of the
better off.
Among WUA members there are also biases in the
pattern of participation. The uneducated are less
likely to participate in meetings, and the better off
are most likely to feel they have a role in decision
making and influencing water management. Lead-
ership of WUAs is heavily concentrated among
higher castes. For these reasons, WUAs can be-
come a channel for elite capture rather than for
empowerment of the poor. Evidence from the
field supports the argument that WUAs can be-
come hostage to local politics, which may mean
elite capture but can also mean capture of proj-
ect benefits by one group or locale at the ex-
pense of others.
Water is not being fairly distributed either within
or between communities, and WUAs have only
limited ability to do anything about this. Many of
these findings reflect similar findings for other
community-based initiatives supported by the
Bank.
WAT E R U S E R A S S O C I AT I O N S
2 7
WUA involvement in O&Mhas shifted to addresspriority problems.
The financial sustain-ability of O&M is doubtfuland may undermine WUAsuccess in that area.
In summary, the Bank has been overly optimistic
about what may be achieved through WUAs ( this
supports the conclusion of IEG’s sector study,
IEG 2002, pp. 43–45), underestimating the need
for continued government involvement. This ap-
plies to technical issues, such as the design of
rehabilitation works, as well as to the ability to
intervene in water disputes. For the time being at
least, it also applies to continued subsidization of
system operation. At the same time, the reasons
for decentralizing WUAs need to be remembered,
in particular, the weaknesses of government irri-
gation departments in efficient management of the
irrigation system. Successful reform needs to ad-
dress these weaknesses with a realistic assess-
ment of what WUAs can and cannot do.
2 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Chapter 4Evaluation Highlights• The impacts of irrigation on crop
yields were strong, but not as signif-icant as appraisal estimates putthem.
• The impact of irrigation on croppingdiversity was minimal, overridden bya strong preference for cultivatingrice.
• Cropped area and cropping intensityincreased, though not as much asappraisal estimates predicted.
• Irrigation increased farm incomesand smoothed income from year toyear.
• Employment of nonhousehold laborincreased with irrigation, benefitingwomen the most.
• The ERR to investments in these proj-ects was negative.
• Risk calculations in Bank appraisalsare too simplistic.
Two men working in an irrigated field in India. Photo by Ray Witlin, courtesy of World Bank Photo Library.
3 1
Economic Benefits
Direct Production ImpactsThese production impacts come from three
sources: higher yields, diversification into higher-
value crops, and increased cropping intensity.2
Higher yieldsAt appraisal, the main income benefits of irrigation
were expected to come from higher yields, which
would result in higher net farm income. For ex-
ample, in the SRSP command below Lower Man-
air Dam (which includes the areas covered by the
survey), the SAR predicted that paddy yields would
rise from 3.2 metric tons per ha without the proj-
ect to 5.4 metric tons with it.3 That is, the SAR as-
sumed a 69 percent increase in paddy yields. This
percentage increase is toward the upper end of im-
pact estimates, with the likely value somewhat
less. But it is also important to note that the “with-
out project” yield is high given actual yields.4 This
matters because it is the absolute increase in yields
that determines the absolute increase in net farm
income, which is the main project benefit.
This study used several sources to estimate the
impact on yields. The IEG survey data were used
to calculate double-difference estimates (see ap-
pendix C).5 These data came only from the reach
below Lower Manair Dam, so additional data from
the Implementation Completion Report (ICR),
SRSP baseline data, and data collected by IEG
from various official sources in Andhra Pradesh
were used to make impact estimates for the higher
reaches,6 though these estimates are single dif-
ferences. These two approaches give consistent
results of an increase in paddy yields of 1–1.5
tons per ha, that is, around 20 percent–50 per-
cent.7 The yield increases for turmeric and chilies
are less in absolute terms but in a similar range
in percentage terms. Cotton has a larger per-
centage increase of around 80 percent and is the
only crop for which the impact on yields appears
to come close to that assumed at appraisal.
An alternative approach is to simply
compare crop yields for the same year
on plots without irrigation with those
with access to canal or tank water. This
analysis was done for paddy in 2006.8
The results show that the impact of ir-
rigation on yield was relatively small in
the 2005–06 kharif season (which is to be ex-
pected, because the monsoon was good) but
larger in the rabi season. Averaging over these
results gives a similar result to that found in the
double-difference analysis.9 Finally, the “natural
experiment” of comparing the year in which
no water was released in SRSP with earlier years
gives an impact on rice yields of just 14 percent
in the kharif season but 66 percent in the rabi
season.
The analysis offers two clear findings. First, access
to irrigation has a significant impact on yields
(figure 4.1). But, second, these increases are less
than those assumed at appraisal, ranging from
one-third for the high-value chilies and turmeric
(34 and 37 percent, respectively) to just under
three-quarters (72 percent) for cotton. For paddy,
Direct production benefits accrue to farm households accessing irri-
gation. Once the SRBC project is completed, the investments un-
dertaken by AP II and AP III will provide these direct benefits to
approximately 212,000 households.1
Estimates of impact onyield found a consistent20 percent–50 percentincrease—substantial, butwell below appraisalestimates.
the most important crop, the assumed
impact at appraisal was almost twice
what has actually been realized.10 Com-
pared with appraisal expectations, these
shortfalls in yields are the norm rather
than the exception in irrigation proj-
ects.11 Because no data are available for the SRBC
command (few farmers have been connected to
the canal system), similar reductions in expected
yield are applied to that case (see appendix C).
Assuming everything else remained as assumed
in the appraisal report, then halving yields com-
pared with their expected values reduces the ERR
in the SRSP (below Lower Manair Dam) area from
the 21 percent calculated at appraisal to 12 per-
cent.12 Applying the same 50 percent reduction
in yields to the whole project reduces the over-
all ERR from 24 percent to 8 percent.13
DiversificationIrrigation is intended to promote agricultural di-
versification, which helps reduce poverty by in-
troducing higher-value crops. Also, a diversified
production portfolio is more resilient to nega-
tive external shocks. The economic analysis in
the SARs for the two projects assumes a diversi-
fied production structure on farms receiving canal
irrigation, notably the production of a broader
range of high-value cash crops, although it is
stated that agricultural support services are nec-
essary to attain diversification.
However, the survey data show that households
with irrigated plots have less diversified cropping
patterns, and farms with access to canal and tank
irrigation are less diversified than those depend-
ing on boreholes and tubewells. The lower level
of diversification reflects the greater reliance on
rice by those with access to irrigation schemes.
These findings can be shown in several ways.
Figure 4.2 shows the share of cropped area, clas-
sifying households by the type of irrigation to
which they have access. The figure also shows the
crop shares that the Bank assumed would be
adopted in project areas, with clearly consider-
ably more diversification being assumed than
has been realized. Those with tank and canal ir-
rigation devote the highest share (55 percent of
gross cropped area) to paddy, 6 percent more
than do farmers without irrigation. The share of
rice in total crop income in 2006 was about 15 per-
cent greater for farmers with canal or tank irri-
gation and more than 20 percent higher than
the year before (appendix table D.9).14 Single-
difference estimates using the IEG data put the
share of paddy in areas with canal irrigation at
up to 30 percent more than in other areas, and
double-difference estimates suggest this increase
is 10 percent to 20 percent. Data from the ICR
show the share of paddy in gross cropped area
up to 39 percent higher in irrigated areas than in
areas without irrigation.
These figures partly reflect differences in the
share of households that depend exclusively on
rice. Compared with other households, twice as
many households with access to canal and tank
irrigation cultivate paddy exclusively (47 percent
in 2005 and 37 percent in 2006). Other than rice,
canal- and tank-irrigated farmers mostly grow cot-
ton; they do not diversify into groundnut, maize,
or high-value crops.
The lack of diversification is not an especially sur-
prising outcome because Andhra Pradesh farm-
3 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Figure 4.1: Irrigation Increases Yields . . . But by Lessthan Assumed at Appraisal (change in yield, tons/ha)
0.0
0.5
1.0
1.5
2.0
2.5
Paddy Cotton Chilies Turmeric
Chan
ge in
yie
ld (t
onne
s/ha
)
Appraisal estimate
Actual impact
Source: IEG survey.
Households with irrigatedplots were found to haveless diversified cropping
patterns.
ers are known to have a strong preference for cul-
tivating rice. Explanations for this preference
range from its reputation as an “easy crop” to
food security and the fact that it enjoys a gov-
ernment-supported price, whereas other crops
may experience much more price volatility.15
Moreover, if a farmer chooses a wet crop, espe-
cially paddy, then farmers in neighboring fields
have little choice but to follow suit, as water leak-
age means their own soil will be unsuitable for
other crops. What is surprising is not the lack of
diversification, but the extent to which it was be-
lieved at appraisal that farmers would diversify de-
spite strong evidence to the contrary. That
evidence has included repeated failure of the
government to enforce production of less water-
intensive crops in the upper reaches of irrigation
systems. The failure to diversify is unique nei-
ther to this project nor to Andhra Pradesh (see,
for example, Plusquellec 2002).
Diversification was expected to result from ex-
tension services, but the data show that few farm-
ers have received such services (ap-
pendix table D.10). The main area of
advice has concerned seeds and use of
fertilizer, but this has still reached only
about 7 percent of households and has
come mostly from community workers (most
likely through the Restoration of Livelihoods Pro-
gram or the Indira Kranti Patham Program) rather
than government extension officers. Construc-
tion delays meant that the support the projects
gave to extension services took place long be-
fore farmers received water, so that support would
have been of limited relevance. But even then such
services were limited: the intended small budget
share for agricultural support services at just 2 per-
cent was squeezed out, falling to just 0.6 percent.
The claims that effective extension ser-
vices help support higher yields and di-
versification may be true, but this
cannot be tested in the virtual absence
of such services in the areas surveyed.
The lack of diversification may not matter much
to those receiving a direct income benefit from
E C O N O M I C B E N E F I T S
3 3
Figure 4.2: Cropping Pattern by Irrigation Source, 2006 (share of gross cropped area)
0
20
40
60
80
100
Canal and tank
Borehole Tubewell Other SAR "with project”
Irrigation Source
New canal New tank
Shar
e of
gro
ss c
ropp
ed a
rea
(%)
Paddy Maize Groundnut Chilies and turmeric Other cropsCotton
Source: IEG survey.
The data confirm a knownstrong preference amongAndhra Pradesh farmersfor cultivating rice.
Although extensionservices were intended tohelp, few farmers receivedsuch services.
irrigation, because rice generates a reasonable
and safe net income and provides food security.
But the additional water requirements of paddy
production have adverse implications for two
other groups of cultivators, most likely to be the
poorer cultivators.
First, the system is designed to provide adequate
water for the cultivation of irrigated dry crops,
such as groundnuts, cotton, and maize. Areas
can be declared as reserved for irrigated dry
crops. However, as the results show, the policy
is not enforced. Canal water charges are on an
area basis rather than volumetric basis, and col-
lection of even those charges is weak, so farm-
ers have no incentive to change cropping
patterns. Indeed, rice is a high-margin crop, pro-
viding a more stable income because the gov-
ernment guarantees the price. Hence, the lower
reaches of the command will only re-
ceive water in very good years, result-
ing in the shortages and disputes
described in the previous chapter and
reducing project benefits.
The second group of losers comprises house-
holds dependent on rain-fed agriculture. The fail-
ure of irrigated farmers to switch to high-value
cash crops means they remain as competitors to
rain-fed farmers, who suffer from lower prices as
a result. But landless households and marginal
farmers who are net food buyers benefit from
these lower food prices. Within the survey area,
just over 40 percent of households (including
noncultivators) do not grow rice at all; another 10
percent grow insufficient amounts to meet their
own consumption needs and so are net buyers.
Thus, half the households will gain from the
irrigation-induced price reduction, although the
result might instead be fiscal savings.16
Cropped area and cropping intensityIrrigation increases the area planted
with crops both by allowing cultiva-
tion of previously uncultivated areas
and by increasing cropping intensity.
Cropping intensity can be increased
by adding growing seasons or by multicropping
(planting more than one crop on the same plot
at the same time, although this is not common on
irrigated land). The area of actual land cultivated
is the net cultivated area; gross cultivated area
refers to the total with double counting (or more)
of land that is multicropped or cultivated in more
than one season.
The net area cropped by the households sur-
veyed increased by nearly 10 percent between
2005 and 2006; cropping intensity increased by
nearly 4 percent, adding up to a 13 percent in-
crease in gross cropped area (appendix table
D.11). Different approaches to assessing project
impact in the SRSP area give estimates of an in-
crease in cropping intensity from 10 percent to
23 percent (appendix table C.18).17
The increase in cropping intensity comes from
both greater intensity of individual irrigation
sources and the adoption of new irrigation sources
that have higher cropping intensity. The main
shift is away from unirrigated plots with low crop-
ping intensity toward higher-intensity tank irri-
gation (figure 4.3).
Canal and tank irrigation are associated with crop-
ping intensities of 200 and 150 percent, respec-
tively (appendix table D.12), an increase that
comes from the possibility of a second, or even
third, crop per year. Double-difference estimates
suggest that canal irrigation increases cropping in-
tensity by 40 percent and tank irrigation increases
it by 20 percent. This average increase of around
30 percent is somewhat above the single-
difference estimates from elsewhere in SRSP and
at the upper end from experience elsewhere.
However, it is the number of sources rather than
the specific source that is the primary determinant
of cropping intensity.18
The survey data show that the increase in crop-
ping intensity comes from the increase in the
number of growing seasons rather than multi-
cropping (figure 4.4). The survey found a 17 per-
cent increase in the number of crops being grown
(separate crops reported on the plot-by-plot col-
lection of production data). This increase came en-
tirely from the expansion of production beyond
the kharif season: in 2005, 96 percent of all crops
3 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
The increase in paddycultivation adversely
affects farmers whocultivate other crops.
The increased croppingintensity was from an
increase in the number ofgrowing seasons rather
than from multicropping.
were grown in the kharif season, compared with
71 percent in 2006. There was no increase in mul-
ticropping between the years. Double-difference
estimates suggest that canal irrigation raises the
number of crops grown on a plot by 0.5, and tank
irrigation increases that number by 0.15 (corre-
sponding to increases in cropping intensity of 38
percent and 12 percent, respectively).
Direct Income Effects: Higher Farm IncomeOutput effects are estimated by calculating the
double-difference value of farm output using
2005 prices for both 2005 and 2006 (see appen-
dix C). The results show that access to irrigation
increases the constant price gross value of out-
put, although the estimates span a large range.
Access to canal irrigation is found to increase
gross revenue by 20 percent, and tank irrigation
increases it by 5 percent.
Farm income is affected by the changing level and
composition of output, as well as by changing
input and output prices. In 2005, at the height of
the drought, the rice price was high; it fell be-
tween 2005 and 2006. However, prices of cash
crops rose between survey rounds (see appendix
table D.13). This study does not attempt to attrib-
ute these price changes directly to the project.
In general, however, the increase in paddy pro-
duction, as well as increased labor demand, will
decrease margins, partly offsetting the yield ben-
efits from irrigation. But given the substantial rise
in cash crop prices, notably those of turmeric
and chilies, the double-difference impact of irri-
gation on incomes appears substantial, reaching
75 percent for canal irrigation.
The farm model finds that irrigation increases
net farm income by 110 percent. About half of this
increase comes from increased cropping intensity
and most of the remainder from higher yields, with
only a small part attributable to changes in the
crop mix. However, these increases are smaller
than those anticipated in the SAR, which esti-
mated that income would more than double in
areas where canals were rehabilitated and nearly
triple in newly irrigated areas.
In general, irrigation, especially large-scale
schemes that are not dependent on the local wa-
tershed for water, guarantees water supply and
so smooths annual variations in production. Crop
E C O N O M I C B E N E F I T S
3 5
Figure 4.3: Cropping Intensity by Number ofIrrigation Sources, 2006
0
40
20
60
80
100
120
140
160
180
200
0 1
Number of irrigation sources
2
Crop
ping
inte
nsity
(%)
Source: IEG survey.
Figure 4.4: Increased Cropping Intensity Comesfrom Additional Growing Seasons
0
100
200
300
400
500
600
700
800
900
Kharif Rabi
2005 2006
Other Kharif Rabi Other
First cropSecond crop
Third crop
Fourth crop
Source: IEG survey.
failures should occur less on irrigated
land than on unirrigated. However, be-
cause of the prolonged drought, in the
2004–05 season no water was released
in the SRSP command for the first time since it
opened. Falling groundwater levels meant that
local irrigation sources were also either dry or that
they provided less water than usual. In 2005, 10
percent of cultivating households got no crop in-
come at all (compared with less than 0.5 percent
in 2006), and the crop failed on 10 percent of all
plots.
The rise of farmer suicides in the state
has been linked to crop failure. In many
instances, farmers could not repay
loans they took to construct a tube-
well, which then ran dry. But unirri-
gated plots were four times more likely to
experience crop failure, except compared with
tank irrigation (table 4.1). In 2006 there were vir-
tually no crop failures on irrigated land. In con-
trast, on unirrigated land, crops failed on 5 percent
of plots. The data do not allow further investiga-
tion of this issue, but evening out crop income is
a substantial benefit from irrigation.
Farm-level returns from irrigationThe returns to the farmer using canal and tank
irrigation are high, as there are substantial bene-
fits and very low costs. The investment costs for
higher yields and cropping intensity are low, as the
farmer is responsible only for the field canals,
which distribute the canal water around the house-
hold’s fields. Moreover, canal water is normally re-
leased into the field by a free intake
(flooding), so most farmers can avoid
pumping costs (equipment and run-
ning costs).19 The recurrent costs as-
sociated with canal and tank irrigation
are the water fees and contribution of
labor to O&M undertaken by the WUA.
Even after the rate increase, the water
fee is only a fraction (less than 3 per-
cent) of the value of the incremental
farm income from gaining access to ir-
rigation. The fee is often not paid, and
few WUAs organize maintenance work.
The farm model (appendix E) shows an average
61 percent increase in net farm income from
canal and tank irrigation.
Tubewell or borewell irrigation also has a high rate
of return, although lower than from canal or tank
irrigation. With this type of irrigation, the farmer
bears the initial investment cost and the recurrent
costs of pump operation.
These costs are lowered by the state govern-
ment’s policy of a zero tariff for agricultural elec-
tricity connections; this is essentially a regressive
subsidy, as it is enjoyed by the better-off farmers.
When it was introduced, the policy could be de-
fended because many farmers were enduring se-
vere economic hardship from years of drought.
But in normal and good years, the benefits from
irrigation easily cover the actual cost of electric-
ity. Unfortunately, the impact of the policy is sim-
ply to encourage overuse of water, to the
detriment of groundwater levels and hence of
other users.
The fact that canal irrigation has sizeable bene-
fits but low costs—and indeed no initial costs to
farmers other than constructing field canals,
which can be done using household labor dur-
ing slack periods—means that there are poten-
tially great benefits for poorer farm households
that cannot access boreholes or tubewells be-
cause of the high initial costs. Those same high
private benefits make the wells attractive to poor
and non-poor alike, so the non-poor will seek
to assert influence to gain access. This occurs in
two ways. First, head-enders use more water than
allowed so that tail-enders get little or no water.
Second, canal water is diverted to tanks so that
only farmers with access to the existing tank sys-
tem get water.
Indirect Income Effects: Employment GenerationThe indirect benefits of irrigation are commonly
referred to as multiplier effects and are often
claimed to be substantial for irrigation.20 These ef-
fects include increased labor supply and demand
for other inputs.
3 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Irrigation substantiallysmoothed crop income
from year to year.
Well-based irrigation alsohas high returns, though
lower than from canal ortank irrigation.
Table 4.1: Crop Failure by Irrigation Source (% of plots)
Source 2005 2006
Canal 6.1 0.0
Tank 15.6 0.0
Tubewell 5.5 0.0
Borehole 4.9 1.3
None 20.6 5.1Source: IEG survey.
If a macro model, such as a computable general
equilibrium model, is used to calculate these ef-
fects, then all rounds of consumption and pro-
duction multipliers are included. But it is not
clear that these multiplier effects should be in-
cluded in a project-level analysis. Such indirect
benefits exist for all projects, not just irrigation,
and are rarely calculated, certainly not beyond the
second round. Hence, resorting to multiplier ef-
fects to boost the benefit stream will undermine
comparability between irrigation and other in-
vestments. More critically, multiplier calculations
have a risk of double counting. Proper use of
shadow prices is preferred.
Take the example of additional employment. If the
labor demand is fully additional, that is, if the
workers would be otherwise unemployed, then
the appropriate shadow wage rate is zero, and the
wage benefits do get included in the benefit
stream.21 There is a complication only if distrib-
utional weights are being used, as the wage ben-
efit should be weighted by the wage workers’
weight, not that of the farm owner. This report
does not use distributional weights but in this sec-
tion looks at who gets the benefits from irrigation.
The principle spillover benefit of irrigation is ex-
panded employment opportunities for members
of landless and marginal households. Agricultural
employment rose markedly between the two sur-
vey years, from 155 days per household each year
to 223 days (see appendix table D.15). Much of this
increase resulted from the revival of agricultural
fortunes following the 2005 monsoon season.
More detailed analysis is required to determine
how much of the increase can be attributed to
irrigation.
The data show that employment of nonhousehold
labor (from outside the household) increases
with the adoption of irrigation. The double-
difference estimate shows that access to canal ir-
rigation increases employment of nonhousehold
labor by approximately 60–80 days a year. Tank
irrigation appears to increase employment some-
what less, with increases from borehole and tube-
well irrigation falling between the two. It may be
the case that canal irrigation required
additional labor for the construction or
rehabilitation of field canals. Tubewell
and borehole irrigation systems have a
greater number of waterings each sea-
son, which can explain the higher labor demand
with respect to tank irrigation.
It might be expected that the additional labor
demand would be muted by households making
greater use of domestic labor. This is not the
case. The double-difference estimates suggest
that irrigation in fact reduces domestic labor use.
To understand this result, one must understand
the general situation, in which use of domestic
labor fell on average from 336 to 317 days22 be-
tween 2005 and 2006. This decrease may be in part
an income effect, substituting leisure for work as
income rises. The more likely explanation is that
households hire out their labor in response to the
greater labor demand by other households.23
Given the lack of alternative employ-
ment in 2005, domestic labor was ab-
sorbed into the family farm, but not in
an efficient manner: labor productivity
was nearly 20 percent lower in 2005
than in 2006.24 With the good rains improving
agricultural performance in 2006, combined with
new irrigation, there was greater labor demand.
Each household supplied on average 223 days of
agricultural labor in 2006, compared with 155 in
2005. The increase in demand for labor led to an
increase in average wages of 10 percent for women
and 5 percent for men; the greater increase in
women’s wages reflects the disproportionate de-
mand for female labor.25 Women’s wages rose
the most in areas of new irrigation.
Women account for the majority of hired labor
in Indian agriculture, with women’s share typi-
cally ranging between 55 and 66 percent (Upad-
hyay 2004). In the IEG survey, women accounted
for 63 percent of hired agricultural labor (share
of days hired) in 2005 and 64 percent
in 2006 (appendix table D.17). In other
words, close to two-thirds of the em-
ployment benefits accrue to women,
E C O N O M I C B E N E F I T S
3 7
The lower costs to farmersfor canal irrigation havepotentially great benefitsfor the poor.
The survey found thatemployment of nonhouse-hold labor increased withirrigation.
Women gained most from the increase inemployment.
reflecting reliance on female labor for
key tasks in paddy production, in par-
ticular weeding and harvesting. A con-
siderable body of evidence shows that
women’s income has a larger impact on child wel-
fare (health and education) than does men’s in-
come, even allowing for the fact that engaging
in labor removes the primary caregiver from the
home.
The employment benefits are not just more em-
ployment but its distribution across the year.
There is much migration from areas of rain-fed
agriculture (mainly to urban areas) because of
lack of income opportunities in rural areas. In
2005, because there was little production in the
rabi season, the average employment of out-
side labor during the rabi season was just 2
days. In 2006, this figure rose to 46 days (ap-
pendix table D.16). The impact analysis shows
that the largest part of the employment effect in-
deed comes in the rabi season, thus smoothing
the seasonal fluctuation in employment (figure
4.5). Despite the monsoon in July–August 2005
that brought rain, rabi season agricultural em-
ployment was negligible in villages without irri-
gation in 2006—but significant in villages with
new irrigation. These figures confirm that a large
part of the additional employment is an “irriga-
tion effect,” which reduces the need for sea-
sonal migration.
Who Benefits? Poverty Reduction andDistributional Effects of New IrrigationThe income benefits from irrigation accrue in the
form of higher net income for cultivating house-
holds and wage income for those they employ,
which in turn is made up of higher wages and
increased employment. The farm model (ap-
pendix E) produces figures for both of these
benefits for a typical hectare of irrigated land. To
see how these benefits are distributed, the in-
cremental farm income is allocated among wealth
quartiles according to access to irrigation, the in-
come from incremental employment to share
of the marginal employment, and wage increases
across all employees. The results are shown in
figure 4.6.
The results show that, in absolute terms, the ben-
efits from irrigation are skewed toward those who
are better off. Those in the poorest quartile ben-
3 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Figure 4.5: Seasonal Employment Effects
0
20
40
60
80
100
120
Canal Kharif2005 2006
In 2005, none of the farms receivedwater, and fluctuation was similar.
In 2006, irrigated farms, which received water, showed less fluctuation.
Rabi Kharif Rabi
Areas withoutirrigation
Areas with new irrigation in 2006
Tank
Add
ition
al e
mpl
oym
ent c
reat
ed (d
ays)
0
5
15
10
30
35
20
25
40
45
Agr
icul
tura
l em
ploy
men
t per
ha
Kharif
Rabi
(a) Largest share of the employment increase from irrigation comes in the rabi season.
(b) Seasonal employment fluctuation is reduced in irrigated areas.
Source: Appendix C.
Notes: (a) is based on household data and (b) on the village-level comparisons.
Irrigation also smoothedseasonal fluctuations in
employment.
efit least from higher net income because few of
them have land to irrigate. Larger farmers are
concentrated in the top quartile, which captures
the bulk of the higher farm income.
This bias is only partially offset by increased em-
ployment opportunities for two reasons. First,
the incremental wages (additional employment
plus higher wages) account for only a minor
share (25 percent) of the total benefits, most of
which are additional employment rather than a
wage increase.26 Second, those in the poorest
quartile benefited little from expanded employ-
ment, which went mainly to households from
the second and third quartiles. The poorest quar-
tile does not benefit from expanded employ-
ment for a number of reasons, including the fact
that it includes elderly and those suffering from
disabilities who are unable to engage in such
work.
Because the benefits to the poorest quartile are
limited, the impact of irrigation on poverty de-
pends a great deal on where the poverty line is
drawn. If it is set at a level such that only those
in the bottom quartile are poor, then the impact
is small (figure 4.7). But with a higher poverty line,
there is a notable impact: bringing irrigation to
a village results in a one-off reduction in poverty
of 10 percentage points, equal to a 25 percent fall.
As discussed below, the dynamic impact on
poverty as a result of reduced vulnerability will
be greater still. There is, however, little impact on
the poverty gap using either poverty line be-
cause of the limited benefits to the poorest.
Although the top quartile benefits most in ab-
solute terms, in relative terms the largest benefit
goes to the second quartile, members of which ex-
perience income growth of 30 percent. Despite
their low benefit, their already low income means
that the poorest quartile also experiences income
growth of 20 percent, compared with 19 percent
growth in the top two quartiles. These effects
mean that irrigation has just a modest positive im-
pact on distribution. However, another major ef-
fect on distribution comes from the dynamic
effects.
Analysis of the Lorenz curves for in-
come between 2005 and 2006 shows a
dramatic reduction in inequality. This
is because many more households had
zero or negative income in 2005 than in 2006
(figure 4.8a). This change was mainly brought
about by the end of the drought. But it was shown
that irrigation reduces the chance of crop fail-
ure, which pushes otherwise well-off families to
the bottom of the income distribution. This fact
is shown in reverse in figure 4.8b, which plots the
Lorenz curve for 2005 and the distribution of in-
come for 2006 with households ranked by 2005
income. The bottom 10 percent of households in
2005 got 10 percent of income in 2006 and 20 per-
cent of farm income (not shown).
So, although comparing “with” versus “without” ir-
rigation for a normal or good year will show a
modest impact on the income distribution, bad
years will be marked by an extreme
worsening in income inequality, as many
farmers experience negative income.
The presence of irrigation can mute this
effect. This reduction in vulnerability
E C O N O M I C B E N E F I T S
3 9
Figure 4.6: Distribution of Irrigation BenefitsSkewed toward the Better Off
01 2 3 4
1,000
2,000
3,000
4,000
5,000
6,000
7,000
8,000
Wealth quintile
Inco
me
gain
per
hou
seho
ld (R
upee
s)
Incremental wages
Incremental employment
Incremental farm income
Source: IEG data.
The benefits fromirrigation are skewedtoward the better off.
The impact of irrigationon poverty dependsheavily on where thepoverty line is drawn.
has a dynamic effect that is not possible to demon-
strate empirically with IEG’s data. Households sub-
ject to repeated negative shocks become heavily
indebted and deplete their assets, constraining
their ability to undertake productive investments.
Reducing the impact of bad years thus aids asset
accumulation and helps households grow out of
poverty.
The overall impact on intravillage income distri-
bution appears modest and may even be positive,
but the intervillage impact may be less beneficial.
At the simplest level, villages with irrigation grow
faster than those without it, thus creating spatial
inequality. This tendency is reinforced by the fact
that villages in head reaches take excess water,
thus depriving those in the tail end, which are
usually worse off to begin with.27 This tendency
toward spatial inequality is offset (1) to the extent
that those in poorer villages perform labor in
irrigated villages, though this seems limited;
(2) from lower food prices if markets are suffi-
ciently integrated; and (3) by other multiplier
effects (a frequently claimed but little quantified
benefit from irrigation).
Revisiting the ERRA number of factors served to undermine the
economic viability of AP III. First, it was saddled
with the sunk costs of AP II, which need to be
taken into account in assessing the return to the
whole investment in the irrigation scheme. Sec-
ond, AP III was affected by the same cost overruns
and construction delays that had affected AP II.
In addition to these problems, the assumptions
made in the SAR regarding yields and diversifica-
tion were rather optimistic. Combining all these
factors means that the ERR on the money that the
government of Andhra Pradesh and the Bank
have spent in SRSP and SRBC has been consid-
erably less than was expected at appraisal. The re-
vised economic analysis for this study finds an ERR
of just 2 percent, compared with the 19 percent
expected at appraisal (table 4.2).28
4 0
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Figure 4.7: Poverty Impact of Irrigation
Pove
rty
mea
sure
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
Upper poverty line Lower poverty line Upper poverty line Lower poverty line
Poverty headcount Poverty gap
Base case (no irrigation)
With project (irrigation)
Source: IEG data.
A Bank study of irrigation in India (World Bank
2003) points out that there are other nonagri-
cultural benefits to irrigation, such as industrial
uses and domestic purposes (drinking water).
Data are not available to quantify these benefits,
but the study shows that for SRSP, 88 percent of
the canal water is used for irrigation purposes.
Therefore, an approximation including the non-
agricultural benefits inflates the benefit stream by
13 percent.29 This raises the overall ERR from 2 to
3 percent (and from 6 to 7 percent excluding AP
II costs).
The yield estimates used in this study are derived
from a number of sources, with remarkable
consistency between those sources. IEG’s final-
estimates of the yield impact are higher than
those used by the
Bank’s irrigation spe-
cialists in preparing
their final review
of AP III. Nonethe-
less, it might be ar-
gued that, with time,
irrigated farmers will
achieve better yields.
Additional calcula-
tions (table 4.3) show
that even if yields with irrigation are as-
sumed to be 20 percent higher than
the increase already estimated, the
overall ERR is still only 4 percent. It re-
mains lower than 10 percent even once
AP II costs are excluded and additional
yields are combined with the nonagricultural
benefits. Even considering the investments from
AP III alone (that is, ignoring AP II costs as sunk
costs), the ERR does not reach the 10 percent
threshold.
The ERR for SRBC is negative, reaching only
zero even when the costs of AP II are excluded,
and 3 percent under the best possible scenario.
All estimates of the ERR for the two components
(that is, those in the SAR and ICR) have shown
E C O N O M I C B E N E F I T S
4 1
Figure 4.8: Income Distribution Improved between 2005 and 2006
(a) Income Lorenz curves, 2005 and 2006 (b) Lorenz curve for 2005 and concentration curve for 2006
–20
0
0 10 20 30 40 50 60 70 80 90 100
20
40
60
80
100
Cumulative share of population
Cum
ulat
ive
shar
e of
inco
me
20052006
45°
–20
0
0 10 20 30 40 50 60 70 80 90 100
20
40
60
80
100
Cumulative share of population (sorted by 2005 income)Cu
mul
ativ
e sh
are
of in
com
e
20052006
45°
Source: IEG survey.
The revised analysis findsan ERR of only 2 percent,compared with the 19 percent expected at appraisal.
SAR IEG
Excluding ExcludingAll costs AP II ICR All costs AP II
SRBC 9 14 12 –2 0SRSP 31 34 16 5 11Whole project 19 24 15 2 6Source: World Bank data and IEG calculations.
Table 4.2: Estimates of the Economic Rate of Return
this same pattern—of SRBC having a
lower return than SRSP, despite SRSP
serving tail-end users.
Several reasons explain this difference.
First, SRBC was new construction,
whereas SRSP was rehabilitation (the SRSP new
construction component of AP II being dropped
under AP III), and rehabilitation is expected to
yield higher returns than construction. In addi-
tion, SRBC was expensive new construction even
before the cost overruns, because the scheme
will irrigate a relatively small area for the length
of the main canal.30 This difference in perform-
ance between the two components has been
exacerbated by the greater delay surrounding the
SRBC component.
These delays in realizing any benefits in the SRBC
command area mean that any realistic set of as-
sumptions about benefits will not yield a satis-
factory rate of return (unlike SRSP, which nudges
toward an acceptable return as the assumptions
become more generous). Saddled with the cost
overruns from AP II, the aggregate investment in
SRBC never had a chance of achieving economic
viability (figure 4.9).
A major implication of this analysis is
the need to avoid delays and cost over-
runs in major construction projects:
the government of Andhra Pradesh is
taking such steps through the Jalay-
agnam program (see appendix G, at-
tachment 2), which seeks to increase the effi-
ciency of government-financed construction.
A combination of problems rendered the project
unviable, yet the appraisal considered the possi-
bility of cost overruns, delayed benefits, limited di-
versification, and a shortfall in yield gains.
Nonetheless, the appraisal concluded that the
project was robust enough to risk. This is because
it adopted the standard approach to sensitivity
analysis employed in Bank appraisals—of varying
one assumption at a time. The possibility that bad
things may be combined was not considered.31
Using the SAR’s own figures, if all the adverse
shocks considered did occur—which they did,
and more severely than assumed in the SAR’s sen-
sitivity analysis—then the ERR for the whole proj-
ect falls from 24 percent to just 3 percent.32
The sorts of problems encountered are far from
unusual in irrigation projects. The simple step of
calculating the ERR for a situation that combined
all the risks would have cast serious doubt on
the viability of the investment. This was not done,
as Bank procedures are far removed from best
practice (box 4.1).
The problem was exacerbated by some of the
unrealistic assumptions made at appraisal, a prob-
lem highlighted in IEG’s recent irrigation sector
review (IEG 2006b). Ex post returns were gener-
ally found to be lower than those at appraisal be-
cause of overestimation of cropped area, expected
yields, and output prices (IEG 2002).
4 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table 4.3: Alternative Estimates of the Economic Rate of Return
SRSP SRBC Whole project
Excluding Excluding ExcludingAll costs AP II All costs AP II All costs AP II
Base case 5 11 –2 0 2 6
Phased yield increasea 6 13 0 2 4 8
Higher yields 6 14 0 2 4 8
Higher yields + nonagricultural benefits 7 16 0 3 5 9Source: World Bank data and IEG calculations.a. Base yield increases in first two years of irrigation, 10 percent higher in years 3–4, and 20 percent higher in years 5 and on.
The appraisal estimateunderestimated risk
because the Bank’sstandard sensitivity
analysis is too simplistic.
The analysis illustrates theirrigation paradox: high
private rates of returncontributing to povertyreduction, but low ERR.
Resolving the Irrigation ParadoxThe irrigation paradox is well illustrated by this
study. The investments in the two schemes sup-
ported by AP II and AP III have yielded a low ERR.
At the same time, the investments have yielded
very high private rates of return to beneficiary
households. The increments in net farm income
and employment have indeed contributed to
poverty reduction. The smoothing of income
made possible by irrigation has helped prevent
households from moving in and out of poverty be-
cause of the vagaries of weather.
The paradox is also readily understood. Farmers
reap the majority of the benefits but bear very few
of the costs. In effect, then, the projects are an in-
come transfer from state coffers (including World
Bank grant and loan funds) to farmers, supple-
mented by subsidies to electricity supply, extension
workers, and so on. Given that the overall finan-
cial and economic returns are negligible, shifting
all costs to farmers would remove the poverty-
reduction impact of irrigation; indeed, poverty
would be increased. The paradox can only be re-
solved by transforming unviable investments into
viable ones. A number of steps will help achieve this.
First, appraisal methods have to undertake a more
rigorous assessment of risks to screen out po-
tentially unviable investments. Second, the plan-
ning about construction time and costs, including
problems in land acquisition, needs to be realis-
tic, and the planning needs to take steps to tackle
the problems that plague these areas. Third, the
assessment needs to be realistic about yield and
diversification benefits. If it is believed that these
really can be achieved with good quality extension
services, then coordination is required (perhaps
through WUAs) to ensure that farmers receive
the information.
If these steps can be undertaken, then invest-
ments may be identified that yield a suitable ERR
and that will yield sufficient income to farmers so
they can meet O&M costs. But
the subsidy element of the cap-
ital cost will remain. So the ques-
tion needs to be asked: Is this
the best use of resources?
E C O N O M I C B E N E F I T S
4 3
Figure 4.9: Investments in SRBC Had No Chance ofProviding an Acceptable Rate of Return
2.0
1.0
0.0
–1.0
–2.0
–3.0
–4.02000 2002 2004 2006
Year of first benefits
ERR
(%)
2008 2010 2012 201
Cost = 60% of actual
Cost = 80% of actual
Actual costs
The figure shows the ERR for SRBC varying two factors: (1) the extent of cost overruns is dif-ferent for each of the three curves, and (2) the date of first benefits becomes later, moving fromleft to right. Even under the most optimistic scenario the ERR is still less than 3 percent.
Source: World Bank database.
Box 4.1: Improving the Quality ofSensitivity Analysis in BankAppraisals
World Bank appraisal reports conduct deterministicsensitivity analyses, varying just one assumption at atime. Several changes of increasing complexity couldbe made to this approach to improve the quality ofappraisal. The simplest change would be to assume theworst for each assumption considered in the analysis.The ERR is truly robust if it remains acceptable in theface of such analysis. If not, then a more elaborate ap-proach is needed.
The second change would be to assign probabili-ties to the possible outcomes. Once that is done, an ex-pected rate of return can be calculated in astraightforward manner, if it is assumed that the vari-ous risks are independent of one another. If they arenot independent, then the correct approach and bestpractice for sensitivity analysis is to specify the jointdistributions and use Monte Carlo simulations to plotthe distribution of the ERR.
Sources: White 2004; Belli and others 2002.
The paradox can only beresolved by transformingunviable investments intoviable ones.
Building a brick main canal in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.
4 7
Summary and Lessons Learned
Two points need be borne in mind when re-
viewing the findings of this study and drawing
lessons from them. First, the projects were “old-
style” irrigation projects of the sort the Bank is
not currently financing in India. Despite this, the
government of Andhra Pradesh is keen to see an
expansion of the area covered by major irrigation
schemes and has launched a substantial program
of new construction. The Bank’s new guidelines
for investment lending in India show that some
of the lessons from the experience of projects
such as AP II and AP III have been learned. Sec-
ond, the reform process establishing WUAs and
defining their roles in practice is likely to be a
long-term one.
Findings on the Six Evaluation Questions
Did investments in irrigated agricultureincrease production in Andhra Pradesh?Yes. Access to irrigation increases both yields and
cropping intensity. Cropping intensity is increased
through the extension of growing into additional
seasons. It also reduces interyear income fluctu-
ations caused by variations in rainfall.
However, the expected diversification into higher-
value crops has not occurred. In fact, irrigated
farmers are less diversified, as they grow more
paddy than farmers whose land is not irrigated,
resulting in excessive water use in the head
reaches.
What are the impacts of PIM on incomes andpoverty alleviation, and who benefits?Gaining access to irrigation raises net farm in-
come by about 60 percent. Indirect benefits come
as well, from additional wage employment. These
indirect benefits are about one-quarter of the
value of the direct benefits.
The top wealth quartile captures the largest por-
tion of the direct benefits and so enjoys the
largest absolute increase in income. But addi-
tional employment income is mostly captured by
the second quartile, which experiences the largest
percentage income growth. The poorest quartile
gains the least absolute benefit but still enjoys in-
come growth of about 20 percent.
Although the largest absolute gain goes to those
who are better off, irrigation results in little change
in the Gini coefficient. The poverty-reduction im-
pact is sensitive to the poverty line. Using an
upper poverty line, irrigation induces a one-off 25
percent reduction in the poverty headcount.
There will be a larger dynamic impact, as irriga-
tion helps smooth productivity and income over
time, breaking the circle of negative income and
debt that can undermine asset accumulation.
Do investments in irrigation yield a satisfactoryeconomic return? In principle, yes; in this case, no. The ex post
rate of return is only 2 percent. The low level re-
sults from cost overruns, construction delays,
and discrepancies between realized income in-
creases and those expected at appraisal.
Two factors are not allowed for in this analysis, one
positive, one negative. First, the ERR calculation
assumes a steady benefit stream. In fact, irrigation
smoothes production income compared with the
“without project” scenario. But the benefits are
only sustained if the canal system is maintained.
Large-scale irrigation will only be economical if
construction costs can be constrained and benefits
sustained. The Bank’s new guidelines prioritize in-
creasing the efficiency of irrigation departments,
which this study confirms to be a necessary start-
ing point for profitable investments.
Was the Bank’s support for canal irrigationrelevant and appropriate?In principle, yes. In practice, however, the an-
swer is twofold: no for the SRBC component and
yes for rehabilitation of SRSP. Irrigation has proven
potential to increase agricultural production and
reduce poverty. Andhra Pradesh has unrealized ir-
rigation potential, so helping meet this potential
appears entirely consistent with a poverty-
reduction strategy—provided, of course, the proj-
ect delivers an acceptable ERR. SRBC failed
to do this, but other aspects of design were in-
appropriate as well.
Excessive faith has been placed in WUAs as a
means of solving problems of both equity in water
distribution and sustainability. But WUAs lack
mechanisms to enforce equitable distribution,
even if they should wish to do so. The physical de-
sign of the system lacked the means for moni-
toring and enforcing proper water usage (that is,
software reform cannot work if the hardware is
missing), and alternative pricing schemes have not
been explored. Increasing the role of WUAs may
improve sustainability, though the evidence shows
that provision for financing O&M matters more
than which bodies are responsible for the works.
It remains possible, of course, that other invest-
ments may have been even more relevant and
appropriate.
Do WUAs facilitate greater participation of the poor and of disadvantaged groups in watermanagement and access to irrigation?No. The poorest are excluded from WUAs be-
cause they are mostly landless. About half of cul-
tivators in canal and tank irrigation schemes report
that they are members of their local WUA. The bet-
ter off are more likely to be members, partly as ir-
rigated land is more likely to belong to the better
off than land in general. Higher castes dominate
the leadership of most WUAs.
Among members, the better off are more likely to
feel they have influence over water management.
WUAs and distributary committees have not pro-
vided an effective mechanism for resolving in-
tercommunity disputes, so the pattern of excessive
water use by farmers in the head reaches con-
tinues, depriving tail-enders, who are usually
poorer, of water. This might have been addressed
by the project committees, but the irrigation bu-
reaucracy has blocked their creation.
Are WUAs an effective means of providingsustainable O&M?It is too early to say. The positive experience with
rehabilitation to date is due to the availability of
external funds (from the Bank’s APERP) to fi-
nance these works. Some of the signs for future
sustainability are not promising: nonpayment or
underpayment of water fees is widespread; most
WUA members do not feel that O&M is their re-
sponsibility; and the amount of money available
to WUAs for O&M is too small, even if they were
to receive their full allocation.
On Selecting Irrigation InvestmentsThe first lessons regard selection of irrigation in-
vestment. More judicious selection would avoid
those projects most likely to have negative returns.
Many previous “lessons learned” still re-main to be learned. The gap between ERR at
appraisal and that found in this study is large.
This has been a consistent finding in IEG reviews
of irrigation projects. The reasons given here are
the same as the reasons found before: under-
estimation of costs and overestimation of bene-
fits. Specifically—
• Construction delays undermine the eco-nomic viability of investments. Serious
delays that postpone the realization of bene-
fits, such as those experienced during these
projects, reduce the rate of return to the in-
vestment. For large-scale irrigation projects to
appear economically attractive, they must be
able to be constructed on time or in a manner
that allows benefits to be phased in at an ear-
lier stage. The government of Andhra Pradesh
4 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
is taking steps to remedy these problems in the
latest phase of irrigation expansion.
• Unrealistic appraisal estimates increasethe likelihood of financing poor invest-ments. The SAR predicted gains from increased
yields and crop diversification that are far from
those that have been realized; the SAR made
the investment appear far more attractive than
it actually was. The Bank’s new guidelines for
irrigation seek to strengthen appraisal proce-
dures, which should directly address this issue.
• The method of risk assessment used inappraisals understates the actual riskthat the project may not be viable. Ap-
praisal reports consider risks one at a time,
not considering the possibility that more than
one of the assumptions in the analysis may
not be realized. Risk assessment methods that
look more rigorously at possible states of the
world will result in more realistic assessments
of project viability.
• Analyses need to value all benefits andcosts. Existing studies (including this one)
neglect some important benefits. The main
ones are increased groundwater levels and
smoothing consumption. All costs should be
valued as well, including realistic full resettle-
ment costs, adverse environmental impacts
(which are most likely long term), and nega-
tive effects on those deprived of water by irri-
gation elsewhere. The Bank’s new guidelines
explicitly recognize the importance of in-
cluding all costs and benefits in the appraisal
analysis.
On Participatory Irrigation ManagementThe lessons regarding PIM are generally applica-
ble to most forms of participatory (community-
based) development.
• Bottom-up development needs supportfrom the top. The initial burst of WUA activ-
ity reflected strong support from the state gov-
ernment. Once the government changed, there
was less support for the program, so it lacked
the central direction needed to bring about
change in the bureaucracy and to support
WUAs as a pro-poor institution at the grassroots
level.
• PIM may alleviate the problems of large-scale irrigation schemes, but it cannoteliminate them. Several of the problems
encountered in these projects, such as con-
struction delays and weakness of extension
services, will not be solved by the introduc-
tion of WUAs. Moreover, WUAs are not real-
izing a potential benefit, as funds for O&M are
insufficient.
• PIM is not intrinsically pro-poor. Farmers
benefiting from irrigation are among the less
poor, and the better-off farmers are most likely
to participate in WUAs. Hence, WUAs are more
likely to reinforce existing social patterns than
to change them.
• WUAs have limited means to effectivelyresolve water allocation disputes.
On the Impact of IrrigationFinally, there are lessons on the design of irriga-
tion projects.
• Irrigation raises output through higheryields and cropping intensity and stabi-lizes production, all of which contributeto poverty reduction. But irrigation in-vestments constitute a sizeable subsidy tofarmers who are not among the poorestor even the poor.
• The direct income benefits to farmersfrom irrigation provide the potentialfor full recovery of the recurrent costs of the services provided. Current water
charges are a tiny fraction of the net increment
in income that irrigation makes possible. In-
creasing these charges is necessary for sus-
tainability of the system.
• Incentive structures, including waterpricing policy, encourage water wastage.
S U M M A R Y A N D L E S S O N S L E A R N E D
4 9
5 3
The Independent Evaluation Group (IEG) study
design required the sample to be split into three
groups: villages receiving canal irrigation since
before the first-round survey in June 2005 (Group
1); villages receiving canal irrigation for the first
time in the two agricultural seasons prior to the
second-round survey, that is, between rounds one
and two of the survey (Group 2); and villages not
receiving canal irrigation yet, but scheduled to re-
ceive it in the near future (Group 3) (table A.1).
The rationale for this design is as follows. Com-
paring the treatment group (Group 2) with the
second control group (untreated; Group 3) will
allow for a standard double-difference estimation
of project effect. The control was selected from
villages scheduled to receive irrigation in line
with a “pipeline approach,” that is, the assump-
tion that those selected for participation but not
yet treated are more likely to share characteristics
with the treatment group than would a randomly
selected control sample. However, there is a con-
cern that farmers take time to adapt their farm-
ing to the availability of irrigation. Control 1
(Group 1) provides a group that has received at
least two treatments (years of irrigation), allow-
ing single-difference estimates with the untreated
group.
AP II and AP III financed work in two command
areas, Srisailam Right Branch Canal (SRBC) and the
Sriramasager Project (SRSP). The study originally
intended to select the sample from an area served
by the new SRBC canal. However, at the time of
the first round of the survey, a year after Andhra
Pradesh III (AP III) had closed, very little water had
been released (irrigating 7,800 hectares [ha] of a
planned 65,000 ha) in the SRBC command, be-
cause the micro-network was at best 20 percent
complete. Completion of the network was esti-
mated for 2008 (now expected for 2007), so the
prospects of a much larger irrigated area by the
time of the 2006 second round were bleak. The
sample was therefore restricted to the SRSP com-
mand, where the main canal and several distrib-
utaries are being extended.
Specifically, the study focused on areas irrigated
by SRSP between kilometer (km) 234 and km
284. This area falls entirely within Warangal district.
Farmers on these distributaries are at various
stages of receiving water: Some had received
water by the time of the first round; some were
to get it by the second round, and some at some
point later. These characteristics fit with the de-
sired three-group sample design. Hence, the sam-
ple design consisted of three groups of six villages
each: (a) Control 1—water released for the first
time in kharif 2003, (b) Project (treatment group):
water released in kharif 2005, and (c) Control 2:
no water released yet.
In this range, other than distributary canal 31 (DC
31), the new canal construction work is not funded
by the Bank. Moreover, DC 31 is not suitable for
inclusion in the sample, as all works were com-
pleted prior to kharif 2003; that is, all villages on
the canal are in Group 1. However, Bank-funded
rehabilitation of the main SRSP canal is what has
made water releases south of km 234 possible
(see chapter 2). Hence, the sample selection is
APPENDIX A: SAMPLE SELECTION OF IEG IRRIGATION SURVEY
Table A.1: Scheme for Survey Coverage
Round 1 Round 2
Group 1 (Control 1)
Group 2 (treatment)
Group 3 (Control 2)Note: Shading indicates that the group has been exposed to irrigation.
justified on three grounds. First, reaching the
areas served has been made possible by Bank-
funded activities, even if the Bank has not directly
financed the canal extensions. Second, the im-
pact of irrigation in the areas may reasonably be
expected to be similar to those in areas which
were financed by the Bank. Third, a major focus
of this study is the economic justification for in-
vesting in further new irrigation works in the state.
Because access to irrigation is determined by
topology, this analogy to a pipeline approach ap-
pears promising. However, it is known that access
to water differs between the head and tail reaches,1
and evidence from the pilot confirmed the preva-
lence of water theft/inefficient usage in upper
reaches. Unfortunately for the evaluation design,
the coverage of the canal system is being ex-
tended mainly by developing the tail reaches of
the longer distributaries (the longest being DC 38
at 80 km in length). Hence, as shown in table
A.2, potential treatment and control sample vil-
lages are not evenly spread across the various
reaches. Control 1 villages are concentrated in the
head reaches and Control 2 villages in the tails.
The exceptions come because there are some
short canals (DC 40 and 41 at around 10 km in
length each), which can be counted as “all head.”
Moreover, other characteristics, which are very
likely correlated with project outcomes, also vary
systematically between project areas and the po-
tential controls. Broadly speaking, Control 1 vil-
lages are most developed and Control 2 least so,
with the treatment group falling in the middle;
see the various indicators in table A.3. Control 2
also has a much higher share of scheduled tribes,
some villages being close to 100 percent sched-
uled tribe communities.
The initial sample design was to stratify by reach
and select even numbers for each group from
head and tail (leaving out the middle reach),
although the sample size was small for Con-
trol 2, especially for head reaches.2 However,
this procedure did not eliminate the systematic
differences between the three groups, so the
validity of the comparison could be called into
question.
Hence, a matching procedure based on propen-
sity score matching was adopted. A probit re-
gression was estimated using census data of “the
propensity to be irrigated”; that is, villages in the
treatment group were given a value of one for the
dependent variable. Irrigation status for each vil-
lage was taken from a list provided by the Irriga-
tion Department in Warangal district. The
regressors were the variables shown in table A.3.
A population-weighted random sample of 12 proj-
ect villages was then drawn, and the nearest neigh-
bor village (based on propensity score) from the
other two groups identified. The 12 were reduced
5 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table A.2: Distribution of Villages by Reach
Head Middle Tail Unclassified Total
Number
Control 1 55 29 5 4 93
Project 26 71 30 5 132
Control 2 3 1 15 0 19
Percent
Control 1 59 31 5 4 100
Project 20 54 23 4 100
Control 2 16 5 79 0 100Source: Calculated from Andhra Pradesh census data, 2000.
Table A.3: Socioeconomic Characteristics of Projectand Potential Control Villages (group averages)
Control 1 Project Control 2
Literacy (%) 48.3 45.4 39.1
Male literacy (%) 61.0 57.5 48.9
Female literacy (%) 35.0 33.2 29.0
No. of educational institutions 1.00 1.00 0.94
No. of medical institutions 1.10 1.08 0.94
Drinking water 1.00 1.00 0.94
Pucca road 0.74 0.65 0.88
Distance to Warangal (km) 37.1 60.1 84.3
Percent scheduled castes 14.2 15.9 10.9
Percent scheduled tribes 19.1 21.6 35.8
Percent of land cultivated 21.8 19.4 16.0
Percent agriculture laborers 24.7 26.9 30.5
Sex ratio (women/1,000 men) 961 974 912Source: Calculated from Andhra Pradesh census 2000.
to 8 to avoid duplicate matches from Control 2
(but even so only 7 villages were picked from
this group). Inspection of the actual values of the
socioeconomic characteristics was used to re-
move the less typical villages, resulting in a sam-
ple of 18, 6 in each group. This procedure
considerably reduced intergroup variation in
socioeconomic characteristics, and where it re-
mained, it did not follow the systematic pattern
shown in table A.3.
Unfortunately, the sample selected was prob-
lematic; field visits revealed that Irrigation De-
partment data on irrigation status of each village
were not reliable. To rectify this situation, a list was
made of three to four possible matches for each
of the two controls, again using the closest match
based on the propensity scores. The actual situ-
ation in these villages was verified in the field, and
thus the final sample selected. But even then vil-
lages that were due to receive water between
the surveys (that is, to be in the project group)
turned out not to receive water.
There are two possible ways to handle this situa-
tion. The first is to include the tail-end villages in
the treatment group, which would give an im-
pact measure of the “intention to treat.” The sec-
ond approach is to recategorize the tail enders into
the “no water” group (Control 1), so the impact
estimate is that for “treatment of the treated.”
But since tail-end villages are unlike other vil-
lages, this latter approach reintroduces systematic
differences into the pre-existing characteristics
of the sample villages, as clearly shown in table A.4.
However (as argued at greater length in appen-
dix C), we know the direction of bias. Tail-end
villages have less access to markets, less well-
educated populations, and are less visited by
extension workers, so we would expect the im-
pact of irrigation to be less than in the head
reaches. Hence, using impact estimates based on
the head reaches produces an upward bias in
those estimates. In addition, the treatment of the
treated impact estimate of course exceeds that
from intention to treat. Nonetheless, it is the
treatment of the treated estimate for the higher
reaches that is used here, because—despite the
upward bias in impact (that is, we have an upper
bound estimate)—the rate of return to the proj-
ect is low.
A P P E N D I X A : S A M P L E S E L E C T I O N O F I E G I R R I G AT I O N S U R V E Y
5 5
Table A.4: Characteristics of Selected Villages
Group
No Newirrigation irrigation Irrigation
Literacy (%) 39 47 42
Male literacy (%) 50 58 54
Female literacy (%) 28 35 31
No. of educational institutions 1.00 1.00 1.00
No. of medical institutions 1.00 1.00 1.00
Drinking water 1.00 1.00 1.00
Pucca road 0.89 0.17 0.17
Distance to Warangal (km) 71 36 44
Percent scheduled castes 12 19 15
Percent scheduled tribes 37 15 29
Percent of land cultivated 19 17 25
Percent agriculture laborers 30 29 24
Sex ratio (women/men) 967 968 977Source: Calculated from Andhra Pradesh census data, 2000.
5 7
This chapter reports the full regressions for the results discussed in chapters 3 and 4. Estimation pro-
cedures are listed in the title of each table. All data for these regressions are from the IEG survey.
APPENDIX B: REGRESSION RESULTS
Table B.1: Determinants of WUA Membership, 2006 (probit regression)
Cultivating households only All households
Coefficient Coefficient(t-stat) (t-stat)
Caste (ref = scheduled caste)
Scheduled tribes –0.60** –0.33
(–2.17) (–1.31)
Other backward caste –0.43** –0.31
(–1.79) (–1.49)
Other caste –0.70** –0.56*
(–2.02) (–1.70)
Wealth (ref = 1st, poorest quartile)
2 0.98** 1.02***
(2.57) (3.12)
3 1.16** 1.23***
(2.95) (3.70)
4 1.95*** 1.96***
(5.01) (5.70)
Land (acres) 0.01 0.04
(0.21) (1.22)
Education (ref = none)
Primary 0.25 0.18
(1.25) (1.06)
Secondary or higher 0.35 0.21
(1.42) (0.91)
Intercept –1.62*** –1.90
(–3.94) (–5.49)Notes: The regression includes village dummies that are not shown. Estimation drops the three villages in which there are no WUA members. These village dummies are the identifying variables for the selection equation for the two stage estimates reported in tableB.2; t statistics in parentheses.
*Significant at 10 percent.**Significant at 5 percent.
***Significant at 1 percent.
5 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table B.2: Determinants of Participation in WUA Activities, 2006 (probit and ordered probit,adjusted for selection into WUA membership)
Regularly Take part FeelAttended attend Taken in WUA influence Voted Voted
WUA WUA part in decision water for WUA for TCmeeting meeting O&M making management president chair
Wealth quartile (ref category = 1)
2 0.57 0.14 4.71 4.33 4.66 0.21 0.10
–1.36 –0.32 (.) (.) (.) –0.61 –0.29
3 0.321 0.27 4.39 4.02 3.67 0.35 0.34
–0.79 –0.68 (13.06)*** (12.16)*** (7.56)*** –0.96 –0.95
4 0.67 0.55 4.30 3.82 3.97 0.36 0.43
–1.63 –1.26 (14.76)*** (10.07)*** (11.59)*** –0.94 –1.16
Education level (ref category = none)
Primary 0.53 0.69 0.36 0.46 0.18 –0.057 0.02
(2.62)*** (2.70)*** –1.58 (1.78)* –0.62 –0.29 –0.10
Secondary or higher 0.40 0.32 –0.32 –0.29 –0.88 0.08 0.13
–1.47 –0.96 –0.84 –0.79 (1.88)* –0.33 –0.52
Caste (ref category = scheduled caste)
Scheduled tribe –0.52 –0.36 0.65 0.18 –0.21 –0.13 –0.11
–1.55 –0.89 –1.41 –0.29 –0.34 –0.49 –0.45
Other backward caste –0.53 –0.56 0.86 0.16 0.46 –0.13 –0.08
(2.05)** –1.64 (2.35)** –0.34 –1.12 –0.51 –0.34
Other caste –0.69 –0.26 0.83 0.64 0.29 –0.03 0.01
(1.69)* –0.57 –1.64 –1.06 –0.43 –0.08 –0.02
Land (acres) –0.003 0.01 –0.01 0.01 0.01 –0.002 –0.01
–0.14 –0.33 –0.49 –0.27 –0.4 –0.07 –0.18
Lambda –5.343 –3.409 –5.118 –3.975 –3.838 –6.564 –6.104
(5.90)*** (3.63)*** (5.41)*** (3.75)*** (3.87)*** (7.59)*** (7.32)***
Constant 1.787 0.177 –3.414 –3.811 –3.88 3.124 2.733
(2.23)** –0.22 (5.35)*** (4.40)*** (4.95)*** (4.13)*** (3.80)***
Observations 566 566 566 566 566 566 566Note: Sample is WUA members only. Absolute value of t statistics in parentheses. WUA = water user association.
* Significant at 10 percent. ** Significant at 5 percent.
*** Significant at 1 percent.
A P P E N D I X B : R E G R E S S I O N R E S U LT S
5 9
Table B.4: Dependent Variable: Value ofTotal Farm Output
Coefficient t-stat
Log of land (ha) 0.57 13.25***
Log of capital (service value) 0.02 1.89*
Log of draught animals (value) 0.02 2.03**
Log of male days own labor 0.07 2.29**
Log of female days own labor 0.04 3.30***
Log of female days hired labor 0.10 3.10***
Log of male days hired labor 0.05 1.79*
Log of fertilizer (expenditure) 0.09 3.15***
Log of seeds (expenditure) –0.02 –1.07
Log of pesticides (expenditure) –0.01 –0.63
Log of other inputs (expenditure) 0.00 –0.05
Rabi season (dummy) 0.13 2.17**
Summer season (dummy) 0.03 0.34
Permanent crops (dummy) 0.13 1.00
Log of age of head of household 0.06 0.60
Log of education of head of household 0.11 2.53***
Share of canal-irrigated land 0.04 0.25
Share of tank-irrigated land 0.43 3.60***
Share of tubewell-irrigated land 0.30 2.45***
Share of borewell-irrigated land 0.29 2.65***
Crop shock (dummy) –0.19 –2.84***
Log of distance from market (km) –0.01 –0.46
Constant 7.66 17.53***
Observations 506.00
R-square 0.68Note: Value in rupees; ha = hectares.***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.
Table B.3: Determinants of Perception thatWater Is Fairly Distributed within Village(probit using 2006 data)
(1) (2)
Wealth quartile (ref = 1)
Quartile 2 –0.10 –0.11
Quartile 3 0.10 0.01
Quartile 4 0.14 0.01
Education level (ref = none)
Primary 0.07 0.09
Secondary or higher –0.26 –0.22
Land (ref = landless)
Marginal 0.96*** 0.97***
Small 1.09*** 1.14***
Large 1.22*** 1.25***
WUA
Household member 0.76*** 0.80***
Village membership rate –0.48 –1.38**
Irrigation characteristics
Proportion household land irrigated 0.84* . .
Number of waterings of irrigatedplots 0.07*** . .
Village water adequacy . . 0.98
Intercept –2.63*** –2.88***
Number of observations 896 896Note: . . = negligible.***Significant at 10 percent. ***Significant at 5 percent. ***Significant at 1 percent.
6 0
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table B.5: Dependent Variable: Production per Hectare
Coefficient t-stat
Log of land (ha) –0.30 –8.98***
Log of capital (service value) 0.02 2.44**
Log of labor (days) 0.13 3.45***
Log of all inputs (expenditure) 0.15 4.52***
Log of age of head of household –2.22 –1.57
Log of age of head of household squared 0.31 1.63*
Log of education of head of household 0.11 3.64***
Red sandy soil (dummy) 0.11 1.89*
Black soil (dummy) 0.18 3.28***
Chaudu soil (dummy) 0.18 3.12***
Irrigation sources (number) –0.17 –2.66**
Rabi season crop (dummy) –0.32 –5.72***
Summer season crop (dummy) –0.44 –4.77***
Crop shock (dummy) –0.13 –2.94***
Rice crop 1.12 7.55***
Maize crop 0.66 4.24***
Groundnut crop 0.24 1.50
Cotton crop 0.44 3.39***
Chilies crop 0.14 0.57
Turmeric crop 0.69 4.74***
Pulses crop –0.31 –2.09**
Canal-irrigated field (dummy) –0.50 –2.44**
Tank-irrigated field (dummy) 0.29 2.92***
Groundwater-irrigated field (dummy) 0.00 0.02
Canal x rice 0.64 2.92***
Canal x maize 1.09 2.81**
Canal x groundnut 0.52 1.33
Canal x chilies 0.67 1.90*
Canal x turmeric 1.04 2.31**
Groundwater x rice 0.15 1.71*
Groundwater x chilies 0.45 2.11**
Constant 8.95 3.36***
Observations 1,113.00
R-square 0.45Note: Production measured in kilograms; ha = hectares.***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.
A P P E N D I X B : R E G R E S S I O N R E S U LT S
6 1
Table B.6: Dependent Variable: Value of Fertilizer Expenditure
Coefficient t-stat
Land (ha) 94.43 1.39
Canal-irrigated land (ha) 660.11 3.60***
Tank-irrigated land (ha) 692.34 4.65***
Tubewell-irrigated land (ha) 399.42 3.29***
Borewell-irrigated land (ha) 435.32 3.41***
Pumpset (dummy) 180.75 0.43
Rabi season (dummy) –299.47 –1.76*
Summer season (dummy) 355.22 1.39
Red sandy soil (dummy) 186.99 2.66***
Black soil (dummy) 232.92 3.34***
Chaudu soil (dummy) 280.62 3.38***
Extension (dummy) –181.51 –0.21
Female-headed household –1,784.16 –2.11**
Age of head of household 4.49 0.29
Education of head of household 22.17 0.18
Cultivators in household (number) 315.56 1.92*
Laborers in household (number) –345.65 –2.39**
Off-farm nonagricultural workers (number) –722.60 –2.29**
Scheduled tribe –1,550.92 –1.84*
Scheduled caste –1,600.94 –2.16**
Other backward caste –1,037.31 –1.51
Wealth index 2457.82 1.32
Draught animals (number) 144.72 0.81
Sheep (number) 47.32 0.44
Goats (number) 71.92 1.07
Distance to fertilizer shop (km) –33.19 –1.11
Share of maize land –1,950.32 –1.40
Share of groundnut land 1,828.03 1.33
Share of cotton land 617.52 1.05
Share of chilies land 3,526.50 2.90***
Share of turmeric land 3,569.64 2.86***
Share of pulses land –83.87 –0.08
Share of other crops 1,616.61 0.79
Male wage (village) –19.73 –0.22
Constant (village) 6,841.99 1.07
Observations 491.00
R-square 0.56Note: Village level price variables for major crops were included but not shown as insignificant; ha = hectares. ***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.
6 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table B.7: Dependent Variable: Value All Inputs Expenditure
Coefficient t-stat
Land (ha) 43.29 0.30Canal-irrigated land (ha) 2,080.30 5.63***Tank-irrigated land (ha) 1,833.91 6.24***Tubewell-irrigated land (ha) 2,086.32 8.64***Borewell-irrigated land (ha) 1,674.16 6.71***Pumpset (dummy) –74.25 –0.08Rabi season (dummy) –1463.38 –4.46***Summer season (dummy) –485.95 –0.89Red sandy soil (dummy) 422.87 2.84***Black soil (dummy) 530.38 3.57***Chaudu soil (dummy) 848.73 4.78***Extension (dummy) 2,163.85 1.17Female-headed household –3,443.59 –1.95**Age of head of household 24.19 0.73Education of head of household –94.22 –0.35Cultivators in household (number) –86.85 –0.25Laborers in household (number) –302.06 –0.98Off-farm nonagricultural workers (number) –1,437.84 –2.20**Scheduled tribe 326.18 0.19Scheduled caste –305.26 –0.20Other backward caste 1,166.22 0.82Wealth index 8,714.84 2.21**Draught animals (number) –267.84 –0.70Sheep (number) 722.09 3.07***Goats (number) 10.67 0.07Distance to fertilizer shop (km) –44.89 –0.71Share of maize land –271.01 –0.09Share of groundnut land 4,623.78 1.55Share of cotton land 2,468.32 1.99**Share of chilies land 9,236.46 3.52***Share of turmeric land 5,975.32 2.20**Share of pulses land 801.12 0.36Share of other crops 6,347.78 1.44Rice price (village) –5.14 –0.70Maize price (village) –4.58 –1.28Groundnut price (village) 0.71 0.40Cotton price (village) 6.57 0.84Chilies price (village) 1.16 1.14Turmeric price (village) –0.46 –0.54Pulses price (village) –1.09 –1.01Other prices (village) 0.40 0.57Male wage (village) 55.17 0.28Constant (village) –1,0748.02 –0.83Observations 514R-square 0.60Note: Value in rupees.***Significant at 1 percent.**Significant at 5 percent.
A P P E N D I X B : R E G R E S S I O N R E S U LT S
6 3
Table B.8: Total Household Labor Demand (days)
Coefficient t-stat
Land cropped –0.40 –0.06
Canal-irrigated land 58.10 3.42***
Tank-irrigated land 51.02 6.32***
Tubewell-irrigated land 86.34 8.40***
Borewell-irrigated land 40.07 3.63***
Share of maize land –10.93 –0.07
Share of groundnut land 21.12 0.13
Share of cotton land 100.20 1.50
Share of chilies land 120.60 0.92
Share of turmeric land –12.27 –0.09
Share of pulses land –17.92 –0.13
Share of other crops 557.04 2.36**
Share of red sandy soil –172.12 –2.63***
Share of black soil –12.59 –0.22
Share of chaudu soil –85.87 –1.49
Draught animals (number) 48.78 2.56***
Goats (number) –5.96 –0.79
Education of head of household –16.52 –1.14
Female-headed household –137.53 –1.35
Age of head of household 2.13 0.90
Female children (number) –14.37 –0.61
Male adults (number) 11.35 0.40
Female adults (number) 101.85 2.92***
Elderly male (number) –35.25 –0.51
Elderly female (number) –44.65 –0.68
Scheduled tribe 105.56 1.16
Scheduled caste 66.65 0.83
Other backward caste 84.02 1.18
Wealth index 412.22 2.14**
Male wage (village) –3.87 –0.31
Constant 3.50 0.01
Observations 471
R-square 0.52***Significant at 1 percent.**Significant at 5 percent.
6 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table B.9: Dependent Variable: Demand for Own HouseholdLabor (days)
Coefficient t-stat
Land cropped 5.66 1.82*
Canal-irrigated land 5.24 0.66
Tank-irrigated land –4.44 –1.17
Tubewell-irrigated land 1.37 0.29
Borewell-irrigated land –0.78 –0.15
Share of maize land 43.28 0.59
Share of groundnut land 111.81 1.44
Share of cotton land 34.71 1.09
Share of chilies land 102.92 1.67*
Share of turmeric land 35.54 0.56
Share of pulses land –42.55 –0.63
Share of other crops –14.72 –0.13
Share of red sandy soil –148.50 –4.82***
Share of black soil –20.49 –0.75
Share of chaudu soil –95.51 –3.51***
Draught animals (number) 27.37 3.06***
Goats (number) –3.62 –1.03
Education of head of household –12.46 –1.84
Female headed household –17.35 –0.35
Age of head of household 0.57 0.51
Female children (number) 1.11 0.10
Male adults (number) 33.60 2.52**
Female adults (number) 87.59 5.33***
Elderly male (number) 69.30 2.11**
Elderly female (number) –38.46 –1.24
Scheduled tribe 86.04 2.00**
Scheduled caste 80.59 2.12**
Other backward caste 93.99 2.82**
Wealth index 61.59 0.68
Male wage (village) –11.61 –1.95
Constant 438.61 2.05
Observations 464
R-square 0.33***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.
A P P E N D I X B : R E G R E S S I O N R E S U LT S
6 5
Table B.10: Dependent Variable: Hired Labor Demand (days) (Tobit)
Coefficient t-stat
Land cropped –6.29 –0.95
Canal-irrigated land 52.67 3.15***
Tank-irrigated land 55.44 6.90***
Tubewell-irrigated land 85.03 8.35***
Borewell-irrigated land 41.15 3.73***
Share of maize land –71.53 –0.41
Share of groundnut land –92.73 –0.56
Share of cotton land 64.23 0.94
Share of chilies land 8.25 0.06
Share of turmeric land –58.17 –0.43
Share of pulses land 32.24 0.20
Share of other crops 569.70 2.44**
Share of red sandy soil –39.39 –0.59
Share of black soil 4.09 0.07
Share of chaudu soil 5.08 0.09
Draught animals (number) 21.64 1.12
Goats (number) –2.52 –0.31
Education of head of household –5.69 –0.39
Female headed household –112.38 –1.12
Age of head of household 1.56 0.66
Female children (number) –12.63 –0.53
Male adults (number) –21.82 –0.76
Female adults (number) 18.75 0.51
Elderly male (number) –101.37 –1.45
Elderly female (number) –8.52 –0.13
Scheduled tribe 42.35 0.45
Scheduled caste –14.08 –0.17
Other backward caste –10.16 –0.14
Wealth index 358.65 1.83*
Male wage (village) 10.31 0.80
Constant –521.68 –1.12
Observations 435
R-square 0.48***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.
6 7
Single-Difference Estimates fromSecondary SourcesSingle-difference estimates of the impact of irri-
gation on crop production are available by com-
paring irrigated and rain-fed plots in comparable
areas. Geographic proximity is taken to ensure
comparability. Such estimates are the most com-
mon method of estimating the impact of irriga-
tion. Single difference estimates are reported
here from the economic analysis for the AP III Im-
plementation Completion Report (ICR), IEG’s
own reestimation following the ICR methodology,
the AP III Baseline Survey, and the IEG survey data.
The SRSP command covers three districts: Karim-
nagar, Nizamabad, and Warangal. To estimate pro-
duction impacts in SRSP, the ICR for AP III used data
for Karimnagar district, comparing data from the 16
mandals that fall entirely within the SRSP com-
mand with the 22 mandals located outside the
SRSP command area and that do not have any sig-
nificant surface irrigation sources. The SRBC com-
mand falls within Kurnool district, from which data
were used for the five mandals accounting for 90
percent of the command area. Because no irriga-
tion water had been received at the time this report
was prepared, these data are the “without project”
scenario. The “with project” data are assessed based
on ICR mission’s field visits and discussions. The
SRBC estimates used in the ICR are therefore spec-
ulative rather than being based on firm data.
Table C.1 summarizes the results from the ICR’s
analysis regarding yields, and table C.2 those for
cropping patterns and intensity. The years used
for the analysis are not stated.
Tables C.3 and C.4 present IEG’s own calcula-
tions of the same analysis on an annual basis for
the years 2002–03 to 2004–05 regarding crop-
ping patterns and intensity. The data show the fall-
off in production in the last year, the last year of
the drought in which no water was released in the
SRSP command. This year can be seen as a natu-
ral experiment, the fall in yields (table C.5) partly
capturing an “after versus before” irrigation effect
(but also showing the impact of general water
shortage in unirrigated areas).
The IEG results differ somewhat from those in the
ICR. The command area cropping intensity is
similar, but lower in the noncommand compari-
son group mandals. In the IEG data the absolute
share of paddy is lower in both treatment and
comparison groups and the difference smaller. But
the paddy remains more common in irrigated
areas than unirrigated, especially in the kharif sea-
son (paddy has quite a high share of noncommand
area rabi production because rabi production is
necessarily irrigated). Finally, the reported yields
in the ICR are higher than those in the more re-
cent data, though the “natural experiment” of no
SRSP releases in the last year gives very similar
yield impact estimates to those from the SRSP
baseline, which are reported next.
The second set of single-difference estimates is
from the AP III baseline report for SRSP. The yields
for the command area are based on crop-cutting
experiments carried out by the Agriculture De-
partment, and those for noncommand areas in the
three districts are based on the data from the Di-
rectorate of Economics and Statistics. Where these
estimates are not available, SRSP estimates are
compared with the state-level estimates. Table
C.6 reports the yield estimates obtained in this
manner.
A final set of single difference estimates comes from
the IEG survey data, taking the comparison of
APPENDIX C: SINGLE- AND DOUBLE-DIFFERENCE IMPACT ESTIMATES
(text continues on page 71)
6 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table C.1: Single Difference Yield Estimates (kg 000/ha)
SRSP SRBC Percent increase
WOP WP WOP WP SRSP SRBC
Paddy (kharif ) 4.0 4.6 4.5 5.1 15.0 13.3Paddy (rabi ) 4.3 5.1 4.0 5.4 18.6 35.0Maize (kharif ) 3.3 3.9 n.a. n.a. 18.2 n.a.Maize (rabi ) 4.0 5.0 n.a. n.a. 25.0 n.a.Cotton (kharif ) 1.0 1.5 1.0 1.5 50.0 50.0Chilies (kharif ) n.a. 1.5 n.a. 1.5 n.a. n.a.Chilies (rabi ) n.a. 1.5 n.a. 1.5 n.a. n.a.Groundnut (kharif ) n.a. n.a. 1.0 2.0 n.a. 100.0Groundnut (rabi ) 1.1 1.7 1.5 2.5 54.5 66.7Groundnut (kharif ) n.a. n.a. 1.0 2.0 n.a. 100.0Sunflower (kharif ) n.a. n.a. 0.6 1.2 n.a. 100.0Sunflower (rabi ) n.a. n.a. 1.0 2.0 n.a. 100.0Sorghum (kharif ) n.a. n.a. 1.0 2.0 n.a. 100.0Sorghum (rabi ) n.a. n.a. 1.5 2.5 n.a. 66.7Chickpea (rabi ) n.a. n.a. 1.0 2.0 n.a. 100.0Source: Economic and financial analysis working paper for AP III ICR.Note: WOP = without project; WP = with project; n.a. = not applicable.
Table C.2: Single-Difference Cropping Pattern (share ofcropped area) and Intensity
SRSP SRBC
WOP WP WOP WP
Cropping patternsKharif
Paddy 42 70 2 4Sorghum 0 0 5 4Sunflower 0 0 5 15Maize 6 9 0 0Groundnut 0 0 2 22Cotton 12 20 15 5Chilies 0 1 0 5Rain-fed crops 40 0 0 0Total 100 100 29 55
RabiPaddy 21 32 2 1Sunflower 0 0 6 25Sorghum 0 0 25 0Chickpea 0 0 40 35Maize 5 7 0 0Groundnut 5 6 1 29Chilies 0 1 0 5Total 31 46 74 95
Cropping intensity 130 145 103 150Source: Economic and financial analysis working paper for AP III ICR.Note: WOP = without project; WP = with project.
Table C.4: Crop Shares of Cropped Area, by Season
2002–03 2003–04 2004–05
Kharif Rabi Kharif Rabi Kharif Rabi TotalCommand area
Paddy 52 53 46 54 29 21 44
Groundnut 0 11 0 10 0 23 4
Cotton 17 0 19 0 35 0 16
Chilies 3 0 4 0 3 0 2
Turmeric 3 0 3 0 5 0 3
Total 100 100 100 100 100 100 100
Memo items
Total (000 ha) 117 49 119 63 93 31 472
Total as % 2002–03 100 41 102 53 79 27
Noncommand area
Paddy 32 38 17 29 13 25 21
Groundnut 2 16 1 12 1 18 3
Cotton 18 0 14 0 39 0 20
Chilies 3 3 2 4 3 0 3
Turmeric 0 0 0 0 0 0 0
100 100 100 100 100 100 100
Memo items
Total (000 ha) 88 17 117 26 119 17 384
Total as % 2002–03 100 19 133 30 135 20Source: Directorate of Economics and Statistics.Note: ha = hectares.
A P P E N D I X C : S I N G L E - A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S
6 9
Table C.3: Single-Difference Estimate of Cropping Intensity (ratio of gross croppedarea to net area sown)
Table C.5: Karimnagar District Average Yields (kg/ha), 2000–01 through 2004–05
2000–01 2001–02 2002–03 2003–04 2004–05
Kharif Rabi Kharif Rabi Kharif Rabi Kharif Rabi Kharif RabiRice 3.1 3.1 3.0 3.5 2.6 3.2 3.2 3.2 2.3 2.8
Jowar 0.7 1.0 1.0 1.1 1.0 0.8 0.9 0.9 0.9 1.0
Maize 3.2 4.0 3.5 5.0 3.1 4.4 3.9 4.7 2.3 3.5
Groundnut 0.6 1.9 0.6 2.0 0.5 1.6 0.5 2.1 0.9 1.0
Chilies 2.2 2.3 1.9 2.8 1.4 2.5 1.7 3.3 1.4 3.0
Cotton 0.4 0.2 0.3 0.4 0.2Source: Directorate of Economics and Statistics.
2002–03 2003–04 2004–05 Average
Command mandals 141 151 134 142
Noncommand mandals 119 123 116 119
Single difference 22 28 18 23
Kharif RabiNon- Non-
CADA CADA % difference CADA CADA % difference
Paddy
1990–2000 3.1 2.4 27 3.4 2.5 38
1998–99 3.1 2.7 13 3.5 2.9 22
1997–98 2.6 1.9 36 2.9 2.5 18
1996–97 3.0 2.6 14 3.3 2.8 18
1995–96 2.2 2.2 0 2.8 2.6 5
1994–95 2.7 2.5 8 2.9 2.4 21
Maize
1990–2000 3.8 3.3 15 4.2 3.7 13
1998–99 3.7 3.3 11 5.2 3.6 46
1997–98 2.6 1.9 36 3.6 2.7 36
1996–97 3.0 2.6 14 4.8 3.8 26
1995–96 2.2 2.2 0 3.8 3.1 25
1994–95 2.7 2.5 8 3.6 3.2 14
Groundnut
1990–2000 n.a. n.a. n.a. 1.6 1.2 35
1998–99 n.a. n.a. n.a. 1.5 1.5 1
1997–98 n.a. n.a. n.a. 1.1 0.9 17
1996–97 n.a. n.a. n.a. 1.7 1.5 13
1995–96 n.a. n.a. n.a. 1.4 1.0 46
1994–95 n.a. n.a. n.a. 1.6 1.1 43
Single-Difference Yield Estimates, SRSP (kg 000/ha)
Turmeric Chilies Cotton
Percent Percent PercentSRSP State difference SRSP State difference SRSP State difference
1994–95 0.31 0.29 9
1995–96 4.8 4.9 –2 1.6 1.8 –13.0 0.34 0.26 31
1996–97 5.0 4.2 19 2.3 2.2 5.0 0.36 0.32 14
1997–98 5.8 5.7 2 1.6 2.0 –22.0 0.23 0.25 –6
1998–99 5.5 5.6 –2 2.1 2.4 –13.0 0.35 0.20 75
1990–2000 4.2 4.7 –10 1.8 1.8 0 0.34 0.26 30Source: AFC (Hyderabad) AP III baseline survey.Note: CADA = Command Area Development Authority, which includes but is greater than the irrigated area; ha = hectares; kg = kilogram; n.a. = not applicable.
Table C.6: Single-Difference Yield Estimates, SRSP (kg 000/ha)
7 0
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
yields from irrigated plots and unirrigated ones
using the 2006 data. These data were collected only
in Warangal district. In this presentation (table
C.7), “unirrigated” excludes plots irrigated from
groundwater by either tubewells or boreholes.
The impact is largest for paddy in rabi, given the
very low reported yield in the unirrigated plots.
There are three sources of bias in the single-
difference estimates. First, it is assumed that the
“pre-irrigation” values of the indicators were the
same in the treatment and controls. However,
the irrigated areas tend to be more developed than
the unirrigated, having more education, better ac-
cess to markets, and a lower share of short-term
consultants. Hence, these areas may be expected
to have higher yields even in the absence of irri-
gation, so the single-difference estimate has an up-
ward bias.
Second, the estimates assume that irrigated areas
are entirely irrigated and unirrigated areas entirely
not irrigated. In fact, not all the irrigated area
will use canal waters. For this reason, the single-
difference estimate underestimates the impact of
irrigation. Third, the comparison area will in-
clude plots (probably the majority) using ground-
water irrigation. Bringing surface irrigation will
either replace one source with another or allow
conjunctive use of irrigation waters—the impact
of either of these being less than that of moving
from rain-fed to irrigated farming. Thus, the
single-difference estimate is an overestimate of
the likely actual impact of canal irrigation being
introduced. In summary, two of the biases are up-
ward (an overestimate) and one is downward
(an underestimate). (There is also the upward bias
from using the treatment of treated effect; see
appendix A).
By making some assumptions it is possible to
correct the biases. The second bias is most
straightforward, because the additional infor-
mation required is the share of the area that is ir-
rigated in the treatment districts. Data from the
baseline survey put this between 60 and 70 per-
cent; the lower figure is taken here, which gives
the larger increase in yields. Assuming that yields
in the unirrigated part of the treatment group are
the same as those for the comparison group,
then, for example, the 15 percent increase in
kharif paddy yields becomes 25 percent. Deal-
ing with the other two sources of bias requires
further assumptions: (1) that the social and eco-
nomic advantages give a pre-intervention yield
differential of 20 percent and (2) that the bene-
fit of going from groundwater irrigation to hav-
ing access to surface irrigation is 60 percent of the
benefit of going from rain-fed to surface irrigation,
and that this adjustment applies to 50 percent of
irrigated plots (the other 50 percent going from
rain-fed to canal irrigation).
These assumptions are applied in table C.8. The
unadjusted estimate is the average of the ICR es-
timate and the baseline survey estimate (calculated
as an average across all available years). The lower
estimate comes from correcting the sources of up-
ward bias and the upper estimate from correct-
ing that for downward bias. The point estimate
A P P E N D I X C : S I N G L E - A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S
7 1
Table C.7: Plot-Level Single-Difference Estimates from 2006 Survey Data
Single differenceTubewell
Level Kg/ha Percent orUnirrigated Canal Tank Canal Tank Canal Tank borehole
Paddy (kharif) 3.7 3.0 5.6 –0.7 1.9 –19 51 4.6
Paddy (rabi) 1.2 4.9 3.0 3.7 1.8 308 150 2.5
Cotton 1.2 1.2 2.0 0.0 0.8 0 67 1.2
Chilies 0.5 0.7 0.8 0.2 0.3 40 60 1.0Source: IEG survey.Note: ha = hectares; kg = kilogram.
results from applying all three corrections at once,
these adjusted point estimates being slightly
higher than the unadjusted estimates. Consider-
ing the upper and lower estimates, the wide range
in these figures is not surprising. What is most
striking is that for the most important crop—
paddy—the largest estimate of the increased yield
is still less than one-third, and it may be less than
10 percent. These increases are considerably less
than those used at appraisal.
There are two further means to attempt to re-
move the biases in the single difference esti-
mate. The first is to ensure a valid match, that
is, ensure that pre-intervention conditions
were the same and trends similar other than
the presence of the intervention. The second
is to use double differencing to remove the pre-
intervention difference.
Controlling for pre-intervention differences can be
done in two ways. The first is to consider control
at the community level, which was done using the
method of propensity score matching described
in appendix A. As explained in that appendix, the
18 villages were placed in three groups: no canal
irrigation, new canal irrigation between the two
surveys, and existing canal irrigation at the time
of the first round. The tables show the single dif-
ference (changes between the two rounds) for
these groups. The no-irrigation category includes
tail-end villages intended to receive irrigation that
have not actually received it. This approach in
principle ensures that communities have com-
mon characteristics with regard to education,
caste composition, and market access, all of which
may affect agricultural productivity. However, as
seen in appendix A, making the match proved
somewhat difficult, given the fact that villages to
be irrigated are rather different from those al-
ready receiving irrigation. The single-difference
village-level estimates are given by the figures in
the 2006 column headed “Single difference” in
table C.9, which compares 2006 yields in villages
newly connected to the canal system with those
without such access in either year.
7 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table C.8: Single-Difference Yield Impact Estimates after Adjusting for Sources ofBias (percent increase in yield)
Adjusted
Unadjusted Kharif Rabi
Lower Point Upper Lower Point UpperKharif Rabi estimate estimate estimate estimate estimate estimate
Paddy 16 19 8 17 27 9 20 32
Maize 16 26 8 17 27 12 28 43
Cotton 38 n.a. 18 41 63 n.a. n.a. n.a.
Groundnut n.a. 50 n.a. n.a. n.a. 24 53 83Note: n.a. = not applicable.
Table C.9: Community-Level Yields by Project Irrigation Status (kg 000/ha)
2005 2006 Single Change (matched plots)
None New Existing None New Existing difference None New Existing
Paddy (kharif) 2.9 2.5 1.9 3.2 4.9 3.0 1.7 1.2 0.6 1.5
Paddy (rabi) 1.7 1.9 0.7 2.5 4.6 3.7 2.1 –0.7 0.6 1.8
Cotton 0.8 0.8 0.7 1.2 1.8 1.3 0.6 0.1 0.7 0.4
Chilies 0.6 0.4 0.1 1.0 2.0 2.0 1.0 0.5 2.0 0.1Source: IEG survey.Note: ha = hectares; kg = kilogram.
The community-level data can also be used for be-
fore versus after for newly irrigated plots or com-
munities. The average for all plots can be used,
or the sample can be restricted to those growing
that crop in both 2005 and 2006, thus controlling
for ecological conditions (such as soil type). This
is the figure shown in the last three columns of
table C.9, headed “Change (matched plots).”
This “before versus after” approach gives ab-
solute yield increases of 0.7 tons/ha (both sea-
sons), 0.6 tons/ha for cotton, and 2 tons/ha for
chilies; in percentage terms these figures are 24
and 32 percent for paddy in the kharif and rabi
seasons, respectively. They are much higher for
cotton (88 percent) and chilies (500 percent),
though the latter figure is based on too few ob-
servations to be a useful estimate. Before versus
after estimates are overestimates if there is a
general rising trend in yields in nontreatment
areas. Table C.9 shows this clearly was the case,
so these single differences are upper estimates
(and broadly consistent with the upper estimates
shown in table C.8, calculated by adjusting for the
bias in the treatment-comparison group single-
difference data discussed above).
The match can also be improved by controlling
for soil type. Sample sizes are too small to do
this for any crop other than paddy, and the com-
parison has to be for canal and tank irrigation
versus groundwater irrigation. The results, shown
in table C.10, give estimates of increased yields
ranging from 11 to 45 percent.
Measuring Impact Using DoubleDifferencesThe double-difference impact estimate is the dif-
ference in the change in the indicator of interest
between the treatment (project) and untreated
(comparison) areas, or, equivalently, the change
in the difference.
Using the first definition, the single difference is
the before versus after comparison of the indicator
in the project area. Before versus after does not
yield a reliable measure of impact because other
factors will have influenced the outcome during
the intervention. This bias is removed by the dou-
ble difference, which compares the change in
the project area with the change in the compar-
ison area, where the comparison area has been
subject to the same trends, with the important ex-
ception of the intervention being absent.
Using the second definition, the single difference
is the estimate reported in the previous section,
that is, the postintervention difference between
the values of the indicator in the treatment and
A P P E N D I X C : S I N G L E - A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S
7 3
Table C.10: Single Difference Paddy Yields by Soil Type, 2006 (kg 000/ha)
Canal and Single difference
tank Groundwater None kg/ha Percent
All soils
Kharif 5.1 3.6 4.6 1.5 29Rabi 3.5 3.0 2.0 0.5 14
Red sandy soils
Kharif 6.5 3.6 n.a. 2.9 45Rabi 2.8 2.6 n.a. 0.2 7
Red soils
Kharif 5.5 3.0 n.a. 2.5 45Rabi 3.3 2.0 n.a. 1.3 39
Black chouduKharif 4.6 4.1 n.a. 0.5 11Rabi 4.5 3.4 n.a. 1.1 24
Note: ha = hectare; kg = kilogram; n.a. = not applicable.
comparison areas. This approach can be subject
to bias if the preintervention values of the indi-
cator were not the same. Hence, the double dif-
ference takes this into account by comparing
changes between project and comparison groups
rather than levels.
The potential problem with double-difference es-
timates is the presence of selection bias if bene-
ficiaries are selected conditional on variables that
are correlated with the indicator under exami-
nation. If these selection variables are observ-
able, then this bias can be removed by including
them as conditioning variables in the analysis, ei-
ther using a regression approach or matching
observations based on these characteristics. If se-
lection is on unobservables, then the bias cannot
be removed, although various approaches might
still be used to deduce impact by producing upper
or lower estimates if reasonable assumptions
might be made to calculate the magnitude of the
bias, as was done for the single-difference esti-
mates above.
Approach to double-difference estimates in this studyAs described in appendix A, matching for this
study was done at the community level. The
double-difference estimates are constructed us-
ing panel data from the IEG survey at the level of
the plot, the household, or the community. Table
C.11 summarizes the data. The discussion of yields
uses the plot and community-level panels.
The first (or single difference) of the double-
difference estimates can be made in two ways.
The first means of calculating single difference is
to use the 2006 data to compare treatment and
control areas, such as reported in the first section
of this appendix. At the plot level this approach
is the source of the figures in the column headed
“Single difference” in table C.7 and discussed
above. The community-level results were shown
in table C.9. The alternative single difference is the
before versus after, also discussed above.
As described above, the double difference is the
change in the differences, or difference in the
changes. The approach can be biased if the treat-
ment group differs systematically from the con-
trol with respect to variables that are also
correlated with the outcome. But the bias can be
removed by controlling for the selection vari-
ables, which this study does in the selection of
the communities to be included in the control
group. Fixed effects regressions are also used
for some outcomes.
But there is a further complication in making
double-difference estimates for agricultural vari-
7 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table C.11: Classifications According to Irrigation Status
Irrigation status Plot Household Community
None
New
Existing
Plot does not receivecanal or tank irrigationwater
Plot received canal ortank irrigation water forfirst time in kharif 2005
Plot received canal ortank irrigation waterprior to kharif 2005
No plots cultivated by the householdreceive canal or tank irrigation water
At least one plot cultivated by thehousehold received canal or tankirrigation water for first time inkharif 2005
At least one plot owned by thehousehold received canal or tankirrigation water prior to kharif 2005
Community either not connectedto SRPS system, or tail-endvillage officially connected butnot receiving water
Community received water forfirst time in kharif 2005
Community received water priorto kharif 2005
ables, which are subject to annual variations be-
cause of weather conditions. Harvests may be
broadly classified as poor, normal, or good, re-
flecting bad, average, or good years, respectively.
Hence, the year-to-year comparison may cover any
one of nine possible combinations (bad-bad, bad-
good, and so forth).
This creates two problems. First, the impact of ir-
rigation differs between bad and good years; it is
expected to be more in bad years, as large-scale
irrigation ensures water supply, which is less avail-
able to those depending on groundwater sources.
By contrast, in a good year water is plentiful, so
canal access matters less. (These observations
need to be further caveated by season: they apply
to the kharif season, in the survey areas rabi
crops, and summer crops are irrigation depend-
ent). So observations are needed across different
types of year to average the effect. The second
problem is that 2005 was not a bad year but a dis-
astrous one, in which no water was officially re-
leased in the SRSP command. Hence the “already
irrigated” sample did not receive water that year
(though a small number do claim to have done so).
Double-difference estimates
YieldYield estimates were calculated at the plot level,
with plots matched by both plot and season.
Hence, only those plots having the same crop, dur-
ing the same season, in consecutive surveys are
used in the analysis. The project group is identi-
fied in two ways: those plots using a new irriga-
tion source in 2006 (calculated separately for
canal, tank, tube, and borehole) and those plots
using a new irrigation source in 2006 that did not
have access to irrigation in 2005. The comparison
group for the first treatment group consists of
those plots with no change in irrigation source be-
tween 2005 and 2006. The comparison group for
the second treatment group comprises plots with
no irrigation in both 2005 and 2006.
These calculations are made for the four major
crop categories: paddy, cotton, chilies, and
turmeric. The results are shown in table C.12. In
some cases, the sample size of the panel is too
small, so the difference reported is that between
the unmatched group means. In other cases, the
sample size is too small for any estimates to be
made. Hence, although Control 2 is more ap-
propriate, Control 1 has to be used. Table C.13 re-
ports the double-difference estimates of cropping
intensity. The impact of canal irrigation is toward
the upper end of those reported above, though
the impact of tanks is lower and, as explained
elsewhere, tank irrigation picks up some of the
effect of canal water.
A P P E N D I X C : S I N G L E - A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S
7 5
Table C.12: Double Difference Estimates of Impact on Yield (kg 000/ha)
Single difference Double difference
Control 1 Control 2 Control 1 Control 2
Paddy
New canal 1.3 1.4 0.7 1.0
New tank 1.4 1.7 0.7 1.4
New bore 1.0 1.7 0.3 1.8
New tube 1.4 1.6 0.7 1.3
None 0.7 0.3 n.a n.a
Cotton
New canal 0.7a . . 0.7 . .
New tank 0.6 1.0 0.6 0.7
New bore 0.4 0.7 0.4 0.4
New tube 0.5a 1.0 0.5 0.7
None 0.2 0.2 n.a n.a
Chilies
New canal 0.4a . . 0.2 . .
New tank 0.7a . . 0.4 . .
New bore 0.6a 0.7a 0.4 0.0
New tube 1.4a 1.7a 1.2 0.9
None 0.2 0.7a 0.0 0.0
Turmeric
New canal . . . . . . . .
New tank 1.6a . . 1.0 . .
New bore 1.2a . . 0.6 . .
New tube 1.2 . . 0.7 . .
None 0.6 . . 0 . .Source: IEG survey.Note: . . = not calculated owing to small sample size.a. Based on difference in simple (unmatched) group means (because of small panel sample size).
The impact estimates are calculated as follows:
(1) paddy—simple average of canal and tank
double difference using Control 2; (2) cotton—
simple average of tank double difference using
Control 2, and twice the Control 1 estimate for
canal, on the grounds that the Control 2 esti-
mates are approximately twice those of Control
1; and (3) turmeric and cotton—simple average
of canal and tank double difference using Control
1. The choices for making the estimates err on the
side of overestimating the benefits. But as noted
in the main text, estimated impact is on average
only half that expected at appraisal.
It should be mentioned that the yield impact es-
timates in I&CADD’s project completion report
(I&CADD 2004) are higher, averaging around
100 percent for paddy in Karimnagar and Waran-
gal districts; the season is not stated by the date
of data collection (March–May), which suggests
it may be rabi when the impact is higher. How-
ever, the data in that report are suspect because
they also discuss impact in the SRBC command,
7 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table C.14: Double-Difference Estimates of Household-Level Employment Effects
Domestic labor Employment
(days) Kharif (days) Rabi (days) Total
Single difference
New canal Control 1 –107 55 74 113
Control 2 . . . . 91 . .
New tank Control 1 –508 43 106 145
Control 2 –887 45 86 126
New tubewell Control 1 166 40 28 196
Control 2 311 38 34 82
New borehole Control 1 –827 20 30 75
Control 2 –640 10 42 89
None Control 1 –216 24 30 44
Control 2 140 5 19 31
Double difference
New canal Control 1 109 31 44 69
Control 2 . . . . . . . .
New tank Control 1 –292 19 76 101
Control 2 –1027 40 67 95
New tubewell Control 1 382 16 –2 152
Control 2 171 33 15 51
New borehole Control 1 –611 –4 0 31
Control 2 –780 5 23 58Note: . . = not calculated.
Table C.13: Double-Difference Impacton Cropping Intensity
Control 1 Control 2
New canal 41 . .
New tank 14 21
New bore 13 4
New tube 11 16
None 0 0Note: . . = not calculated.
despite the fact that no water had been released
there at the time the report was written. In the
case of paddy yields, the completion report cites
a 53 percent increase in “project villages” in
SRBC. There was no irrigation there, so this in-
crease can be combined with the 100 percent in-
crease in SRSP to give a double-difference yield
impact estimate of approximately 50 percent,
which is more consistent with the estimates used
in this study.
EmploymentDouble-difference employment effects were cal-
culated using the household panel by averaging
across the various irrigation categories (as used
for yields above). These results (table C.14) show
that adoption of canal or tank irrigation increases
employment throughout the year between 80
and 100 days. The fixed-effects regression (table
C.15) gives the same result.
The community-level panel can be used to con-
sider employment from the employee rather
than the employer perspective. The data show
that a new canal gives an increase in household
agricultural employment of about 80 days more
than that with no irrigation (table C.16). The
community-level analysis also shows wage ef-
fects. Wages, especially for women, rose across
the district in 2006, by most in areas with new
canal irrigation (table C.16).
Summary of SRSP EstimatesTable C.17 summarizes the various estimates of the
impact of irrigation on paddy yields. These results
are expressed in percentage terms, but it should
be remembered that it is the absolute increase that
matters for the rate of return analysis, because it
is the absolute increase in net farm income that
constitutes the core of the project benefits (es-
pecially for paddy, which has the largest crop
share). The “without project” value assumed in the
staff appraisal report (SAR) (3.2 tons/ha) is to-
ward the upper end of those observed in the dis-
trict—indeed at levels only recently attained in
the SRSP command itself (figure C.1). Table C.18
provides a summary of estimates for cropping
intensity.
A P P E N D I X C : S I N G L E - A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S
7 7
Table C.15: Fixed-Effects Regression of EmploymentEffects
Kharif Rabi Total
Canal 33.8 27.1 60.9
–1.0 –1.1 –1.3
Tank –22.6 –41.3 –64.0
–1.1 (2.76)*** (2.28)**
Tubewell 1.6 22.5 24.2
–0.1 –1.4 –0.8
Borehole 24.3 48.0 72.4
–1.3 (3.56)*** (2.86)***
Canal* area –2.0 4.8 2.8
–0.6 (2.03)** –0.6
Tank* area 14.3 24.1 38.4
(4.73)*** (10.77)*** (9.17)***
Borehole* area –1.2 –11.8 –13.0
–0.4 (5.41)*** (3.17)***
Tubewell* area 1.6 –9.4 –7.9
–0.5 (4.26)*** (1.90)*
Constant 78.8 18.1 96.8
(6.68)*** (2.08)** (5.94)***
Observations 916 916 916
R-squared 0.32 0.09 0.43
Marginal effect at mean area
Canal 20 61 81
Tank 37 60 97
Tubewell 11 –37 –26
Borehole 9 –21 –12Note: Absolute value of t statistics in parentheses.
*Significant at 10 percent.**Significant at 5 percent.
***Significant at 1 percent.
Table C.16: Change in Employment(days/household/year) and Wages (Rs/day)
Change in Change in wages
employment Female Male
None 33 2.1* 0.2
New 114* 3.3* –1.9
Existing 42* 1.8 2.9*Source: IEG survey.*Significant at 5 percent.
Srisalam Right Bank CanalBecause the SRBC command has not yet received
irrigation, there are no ex post data available for
impact analysis. The approach adopted here is
that appraisal estimates are overly optimistic, so
that lower percentage increases are assumed.
Table C.19 shows the appraisal and ICR estimates,
together with those used in this study. The ini-
tial yields used are usually the lower of the esti-
mates from the two sources. In the case of paddy,
the lower of the two estimates (from the SAR) is
still the highest level achieved in Kurnool district
in the last 15 years; the average from 1989–90
to 2003–04 has been 2.6 tons/ha, peaking at 3.2
tons/ha in 2000–01 (data from the Directory of
Economics and Satistics, Andhra Pradesh).
7 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table C.17: Summary of Impact on Rice Yields (percentage increase)
Kharif RabiSingle Double Single Double
difference difference difference difference
ICR 15 n.a. 19 n.a.
IEG Mandal-level 63 n.a. 37 n.a.
“Natural experiment” 63 14 34 66
Community level 53 84 84 135
Community level (matched plots) 29 –24 14 n.a.
Plot-level 32 24 46 33
AFC (SRSP baseline) 16 n.a. 20 n.a.
Adjusted single difference 17 n.a. 20 n.a.
Memo: assumed in SAR 69 n.a. 69 n.a.Notes: ICE = Implementation Completion Report; n.a. = not applicable; SAR = Staff Appraisal Report.
Figure C.1: Yields in the SRSP Command Area
500
01976–77 1980–81 1984–85 1988–89 1992–93 1996–97 2000–01
Kharif
Rabi
2004–05
1,000
1,500
2,000
2,500
3,000
3,500
4,000
Yiel
d (k
g/ha
)
Years
Source: I&CADD and Directorate of Economics and Statistics.
A P P E N D I X C : S I N G L E A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S
7 9
Table C.19: SRBC Yield Impact Estimates
SAR ICR This study Percentage change
ThisWOP WP WOP WO WOP WP SAR ICR study
KharifPaddy 3.2 5.4 4.5 5.1 3.2 3.8 69 13 20
Cotton 1.1 2.5 1.0 1.5 1 1.5 127 50 50
Sorghum 1.2 2.5 1.0 2.0 1 1.6 108 100 60
Sunflower 0.6 1 0.6 1.2 0.5 0.8 67 100 67
Groundnut 1 2.2 1.0 2.0 1 1.6 120 100 60
Chilies 0.9 2.5 n.a 1.5 0.8 1.6 178 n.a. 60
RabiPaddy n.a. n.a. 4.0 5.5 3.6 5.0 n.a. 38 40
Groundnut 1.0 2.5 1.5 2.5 1.0 1.7 150 67 67
Sorghum 1.2 3 1.5 n.a. 1.2 2.0 150 n.a 67
Sunflower 0.6 2 1.0 1.8 0.6 1.1 233 80 80
Chickpea 0.6 1.5 1.0 2.0 0.6 1.1 150 100 80Source: IEG data.Notes: n.a. = not applicable; SAR = Staff Appraisal Report; WOP = without project; WP = with project.
Table C.18: Estimates of Impact on Cropping Intensity
Single difference Double difference
ICR 15 n.a.
IEG Mandal level 23 n.a.
“Natural experiment” 17 10
Plot level 14 21Note: n.a. = not applicable.
8 1
APPENDIX D: DATA TABLES
Table D.1: Irrigation Spending and Potential Created in Andhra Pradesh
Amount spent Irrigation potential created(million Rs.) (million ha)
Major and Major andmedium Minor Total medium Minor Total
Preplan period 1951 — — — 1.33 1.37 2.70
First plan (1951–56) 375 35 410 0.08 0.03 0.10
Second plan (1956–61) 574 44 618 0.18 0.02 0.20
Third Plan (1961–66) 915 186 1,101 0.37 0.05 0.42
Three annual plans (1966–69) 609 108 717 0.08 0.04 0.12
Fourth plan (1969–74) 1,187 182 1,369 0.19 0.06 0.25
Fifth plan (1974–78) 2,691 388 3,079 0.21 0.09 0.31
Two annual plans(1978–80) 2,577 238 2,815 0.15 0.06 0.21
Sixth plan (1980–85) 7,296 411 7,707 0.31 0.08 0.39
Seventh plan (1985–90) 13,064 1,314 14,378 0.09 0.07 0.16
Annual plan (1990–91) 2,828 478 3,305 0.01 0.01 0.02
Annual plan (1991–92) 3,339 486 3,825 0.01 0.01 0.02
Eighth plan (1992–97) 22,653 1,869 24,522 0.04 0.04 0.08
Ninth plan (1997–2002) 50,272 7,757 58,029 0.58 0.03 116,058.61
Tenth plan (2002–07) 91,538 16,072 107,610 n.a. n.a. 0.94Note: n.a. = not applicable; — = not available.
8 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table D.2: SRSP: Actual Area Irrigated as a Percentage of Planned Area Irrigated
1995–96 1996–97 1997–98 1998–99 1999–2000
Above Lower Manair Dam
Kharif
Head 129 94 140 87 85
Middle 55 74 78 49 85
End 45 40 68 88 88
Rabi
Head 55 76 . . 51 42
Middle 52 57 . . 66 n.a.
End 50 52 . . 67 99
Below Lower Manair Dam
Kharif
Head 25 44 59 66 73
Middle 42 38 55 113 110
End 50 51 53 97 89
Rabi
Head 26 41 . . 55 94
Middle 45 44 . . 98 112
End 40 42 . . 76 n.a
Simple average 51 54 75 76 88Source: AFC SRSP baseline survey, pp. 127–29.Note: . . = no water released this season; n.a. = not applicable.
Table D.3: Area under Different IrrigationSources as Percent of Total Area
2005 2006
Canal 4.6 5.2
Tank 7.8 20.5
Tube 44.8 50.5
Bore 27.0 26.5
None 23.6 11.3Source: IEG survey.Note: Sums to more than 100, as some plots have multiple sources.
A P P E N D I X D : D ATA TA B L E S
8 3
Table D.4: WUA Membership by Socioeconomic Characteristics (% households with WUA member), 2006
% households with member
CultivatingCategory All households households only
Landowning Landless 0.0 0.0
Marginal 18.4 18.4
Small 12.2 12.2
Large 28.2 28.2
Education None 8.4 13.7
Primary 20.7 30.1
Higher than primary 15.6 24.7
Caste Scheduled caste 13.3 28.5
Scheduled tribe 5.0 6.7
Backward caste 13.2 21.8
Other caste 20.5 28.0
Wealth 1 0.8 2.0
2 7.6 11.4
3 9.3 13.7
4 28.0 37.0Source: IEG survey.
Table D.5: Gini Coefficients for Land Ownershipby Irrigation Source
Canal Tubewelland tank and borehole All land
2005 0.57 0.40 0.35
2006 0.45 0.39 0.34Source: IEG survey.
Staff appraisal report
Handbook of statistics Without WithCrop (2003–04) project project IEG survey
Paddy 2.2 3.2 5.4 3.8
Cotton 2.4 1.1 2.5 1.9
Chilies n.a. 0.9 2.5 1.6
Turmeric n.a. 3.2 4.0 2.0Sources: Handbook of Statistics Warangal District 2003–04; AP III ICR, appendix 10.Note: ha = hectare; kg = kilogram; n.a. = not applicable.
Canal Tank Tubewell Borehole All land
2005
1 4 10 7 7 8
2 5 10 23 14 19
3 23 9 28 24 26
4 69 71 43 54 47
2006
1 13 6 4 7 8
2 14 16 16 22 22
3 43 12 17 27 24
4 31 65 63 44 47
Total 100 100 100 100 100Source: IEG survey.
8 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table D.6: Share of Irrigated Land Owned by Wealth Quartile
Table D.7: Involvement in WUA Activities by Wealth Quartile (percent), 2006
Regularly attend Take part in Take part in InfluenceQuartile Member meetings O&M decision making water management
1 0.8 0.2 0.0 0.0 0.0
2 7.6 1.1 3.6 1.7 3.1
3 9.3 1.8 3.2 1.3 0.6
4 28.0 11.0 11.1 4.7 4.4Source: IEG survey.
Table D.8: Actual and Expected Crop Yields (kg 000/ha)
A P P E N D I X D : D ATA TA B L E S
8 5
Table D.9: Crops Shares in Crop Income (percent)
% exclusiveRice Cotton Maize Chilies Turmeric Groundnut rice producers
2005
Irrigation source
Canal 61.7 24.7 0.5 6.8 2.1 2.7 42.1
Tank 61.2 16.8 6.1 2.2 2.9 0.8 48.1
Bore 37.9 36.5 2.9 8.5 5.3 2.7 17.4
Tube 42.1 36.2 0.9 6.3 4.2 3.1 20.0
None 32.1 32.6 1.7 6.5 3.5 1.2 22.0
Number of sources
0 32.1 32.6 1.7 4.5 0.7 1.2 22.0
1 41.2 35.0 2.2 7.3 4.3 2.7 22.3
2 49.2 27.4 4.9 6.0 6.0 2.4 22.9
Total 39.4 33.8 2.3 6.5 3.5 2.2 22.4
2006
Irrigation source
Canal 59.0 15.4 4.2 10.3 7.5 2.8 32.3
Tank 59.7 25.4 2.0 3.5 4.3 1.9 38.8
Borehole 44.5 30.2 2.0 8.0 6.7 2.2 17.4
Tubewell 42.4 28.9 2.8 8.4 8.0 5.5 18.5
None 43.6 34.2 4.8 1.0 1.0 2.2 20.0
Number of sources
0 43.6 34.2 4.8 1.0 1.0 2.2 20.0
1 48.6 27.7 2.4 7.4 5.0 3.0 23.7
2 47.8 27.1 2.3 7.2 11.1 2.7 21.5
Total 48.1 28.3 2.5 6.8 5.5 2.9Source: IEG survey.Note: Income is sales + imputed value of own consumption.
Cultivated area Gross Cropping(net cropped area) cropped area intensity
2005 1,610 2,414 150
2006 1,760 2,738 156
Percent increase 9.4 13.4 3.7Source: IEG survey.
8 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table D.10: Sources of Information on Agriculture-Related Matters
Received Source from which received information
information (as a percent of those who received information)
(percentage of Friendscultivating and Extension Community
households) family worker Training worker
2005
Use of fertilizer 7.7* 7.8 2.2 13.2* 74.7
Seeds 7.0* 8.5* 0.0 5.2 84.0
Pest management 0.9* 39.7 0.0 22.5 18.9
Soil conservation 1.0* 84.0* 0.0 16.0 0.0
Agricultural marketing 0.2 100.0 0.0 0.0 0.0
Irrigation methods 0.3 0.0 0.0 0.0 0.0
O&M of canals 0.0 n.a. n.a. n.a. n.a.
Water management 0.2 100.0 0.0 0.0 0.0
2006
Use of fertilizer 6.7* 50.4* 0.0 16.5 26.3
Seeds 3.4* 39.3* 0.0 32.0* 18.2
Pest management 1.5* 0.0 0.0 75.1* 0.0
Soil conservation 0.3 0.0 0.0 0.0 0.0
Agricultural marketing 4.1* 64.7* 0.0 6.2 18.7
Irrigation methods 0.4 0.0 0.0 100.0 0.0
O&M of canals 0.0 n.a. n.a. n.a. n.a.
Water management 0.0 n.a. n.a. n.a. n.a.Source: IEG survey.Note: n.a. = cannot be calculated as 0 positive responses to screening question; O&M = operations and maintenance.*Significant at 5 percent.
Table D.11: Gross and Net Cropped Area (hectares), 2005 and 2006
Canal Tank Bore Tube
Free intake 74 80 18 8
Manual pump 0 1 3 4
Electric pump 26 19 78 87
Other 0 0 1 1Source: IEG survey.
A P P E N D I X D : D ATA TA B L E S
8 7
Table D.12: Cropping Intensity by IrrigationSource and Number of Sources
2005 2006
Irrigation source
Canal 202 199
Tank 148 151
Tube 159 168
Bore 143 169
None 139 90
Total 150 156
Number of sources
0 139 90
1 150 161
2 180 182Source: IEG survey.
Table D.14: Means of Water Transfer to Field by Irrigation Source, 2006
Table D.13: Average Crop Prices, 2005 and 2006 (Rs per quintal)
2005 2006
Mean Median Mean Median
Rice 718 672 605 581
Maize 929 643 511 500
Groundnut 1,063 977 1,420 1,447
Cotton 1,692 1,714 1,797 1,798
Chilies 1,430 1,441 2,272 2,129
Turmeric 1,011 606 1,605 1,550
Pulses 1,448 1,136 1,632 1,683
Jowar 2,000 2,000 900 900Source: IEG survey.
8 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table D.15: Agricultural Employment (average days per household)
Outside employment Own farm
All Cultivating Noncultivating All CultivatingYear households households households households households
2005 155 57 306 221 336
2006 223 157 313 190 317Source: IEG survey.
Table D.16: Average Employment of OutsideLabor by Season (days)
Season 2005 2006
Kharif 98 145
Rabi 2 46
Total 99 191Source: IEG survey.
Table D.17: Agricultural Employment by Gender (days)
2005 2006 Change
Kharif
Female 63 101 38
Male 37 34 –2
Rabi
Female 1 91 90
Male 1 54 54
Total
Female 64 199 135
Male 38 114 76
Percent female in total 63 64 64Source: IEG survey.Notes: Total is taken as total of subsample means. Differences caused by rounding.
A P P E N D I X D : D ATA TA B L E S
8 9
Table D.18: Agricultural Employment by Wealth and Landholding (average days perhousehold), 2005
All Cultivating All CultivatingQuartile households households Landholding households households
1 278 88 Landless 306 n.a.
2 123 50 Small 79 79
3 159 70 Marginal 22 22
4 62 41 Large 21 21Source: IEG survey.Note: n.a. = not applicable.
Table D.19: Employment of Domestic Labor by Wealth and Landholding (averagedays per household), 2005
All Cultivating All CultivatingQuartile households households Landholding households households
1 134 290 Landless n.a. n.a.
2 227 308 Small 273 273
3 239 348 Marginal 400 400
4 282 364 Large 461 461Source: IEG survey.Note: n.a. = not applicable.
Table D.20: Agricultural Employment by Wealth and Landholding (average days perhousehold), 2006
All Cultivating All CultivatingQuartile households households Landholding households households
1 294 209 Landless 316 n.a.
2 234 182 Marginal 223 210
3 239 168 Small 135 128
4 132 108 Large 66 65Source: IEG survey.Note: n.a. = not applicable.
9 0
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table D.21: Employment of Domestic Labor by Wealth and Landholding (average days per household), 2006
All Cultivators All CultivatorsQuartile households only Landholding households only
1 82 265 Landless 0 n.a.
2 171 269 Marginal 244 265
3 223 338 Small 355 361
4 273 357 Large 383 390Source: IEG survey.Note: n.a. = not applicable.
Table D.22: Agricultural Employment by Wealth and Land Ownership, 2005
Workers as percentage of Days worked as percent of those age 16–65 available days
Outside Outsideemployment Own farm employment Own farm
Wealth quartile
1 50 29 21 9
2 23 55 10 15
3 25 47 10 15
4 11 52 4 18
Landholding
None 57 n.a. 23 n.a.
Marginal 10 67 5 19
Small 3 80 1 23
Large 7 72 2 27
All 27 45 11 14Source: IEG survey.Note: n.a. = not applicable.
A P P E N D I X D : D ATA TA B L E S
9 1
Table D.23: Agricultural Employment by Wealth and Land Ownership, 2006
Days worked as percent of Days worked as percent of available days available days
Outside Outsideemployment Own farm employment Own farm
Wealth quartile
1 73 23 24 7
2 59 41 18 12
3 53 49 17 15
4 29 46 9 19
Landholding
None 68 n.a. 25 n.a.
Marginal 55 59 15 18
Small 38 70 9 24
Large 20 65 6 27
All 53 40 17 14Source: IEG survey.Note: n.a. = not applicable.
Table D.24: Agricultural Day Wage (Rs/day)
2005 2006
Female Male Female Male
Mean 24.95 48.09 27.50 70.72
Median 25 50 30 50Source: IEG survey.Note: n.a. = not applicable.
9 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table D.25: Distribution of Wages by Year by Sex (Rs/day)
Female Male
2005 2006 2005 2006
<=20a 44 13 1 0
25 28 45 0 1
30 26 41 7 16
35 0 0 1 0
40 1 0 3 9
45 0 0 3 0
50 0 1 74 63
55 0 0 2 0
60 0 0 10 4
65 0 0 0 0
70 0 0 0 7Source: IEG survey.a. There is just one female observation below 20 in both years.
Table D.26: Socioeconomic Characteristics by Reach on Distributary Canal
Adult Scheduled Scheduled Pucca Distance toReach literacy caste (%) tribe (%) road (%) Warangal (km)
Head 48.1 15.7 20.1 67 39.1
Middle 45.6 14.1 24.3 74 53.0
Tail 43.6 15.7 18.8 65 76.4Source: Calculated from Andhra Pradesh census.
9 3
The farm model, based on that in the SAR, shows
farm income without project (WOP) and with ir-
rigation (with project, WP) for a typical hectare.
Farm models are estimated separately for SRBC and
SRSP. The basis for the yield increases and crop
shares is explained in appendix C. Farm inputs are
assumed to increase with irrigation, though not by
as much as assumed in the SAR. These increases
in inputs as a result of irrigation are confirmed by
the multivariate analyses (tables B.6–B.10). The
smaller increase than that expected at appraisal can
be linked to the reduced delivery of extension ser-
vices. Lower inputs are one reason that actual
yields are lower than those anticipated at appraisal.
The farm budget is for a typical hectare, with the
net income calculated as a weighted average of net
income from each crop (the weights being the
share in gross cropped area), the total multiplied
by cropping intensity. The resulting net farm in-
comes with and without the project are shown in
table E.1. In the case of SRSP, the IEG estimate lies
between those from the SAR and the ICR. But the
estimated increase in net farm income in the SRBC
command is less than half that calculated by the
other two reports. As explained at various points
in this report, there is no irrigation yet in the SRBC
command. Hence, the ICR relied on “informed
judgments” of the sort used in the SAR, presum-
ably explaining the similar estimate. But the figure
seems somewhat optimistic, being based on yield
increases larger than those observed elsewhere
and a diversification into high-value crops, which
has not occurred in other places (see appendix B).1
The benefits stream in the rate of return analysis
is net farm income (valued at economic rather
than financial prices, though only the latter are
shown here) multiplied by the irrigated area.
Given problems in reaching tail-enders and evi-
dence of overestimation of irrigated area by
I&CCAD, it is assumed that the irrigated area
reaches 80 percent of that stated in the SAR and
ICR. This figure may prove optimistic for SRBC,
where the Bank has questioned the adequacy of
the water supply. The results (table E.2) show
the low return to these investments. Estimates are
shown, including all investment costs for both AP
II and AP III, which is the relevant figure for this
report. Economic rates of return (ERRs) calculated
using only the AP III investment costs are also
shown. Further notes on the calculations will be
found in the tables beginning on page 97.
Poverty and Distributional Impact
Benefits per hectare of irrigationThe income benefits are the increment in farm in-
come and wage income to agricultural employees
(payment to domestic labor is already included in
APPENDIX E: THE FARM MODEL, RATE OF RETURN ANALYSIS,
AND DISTRIBUTIONAL IMPACTS
Table E.1: Net Farm Income (financial prices) With and Without Project (Rs/ha)
SAR ICR IEG estimates Percentage increase
WOP WP WOP WP WOP WP SAR ICR IEG
SRBC 7,469 3,4356 4,894 24,089 5,970 15,526 360 392 160
SRSP 9,796 22,792 8,842 14,965 7,524 16,070 133 69 114Note: ha = hectare; ICR = Implementation Completion Report; Rs = rupees; SAR = Staff Appraisal Report; WOP = without project; WP = with project.
net farm income; see table E.3). The increase in
employment income can be further broken down
into higher wages (benefiting all employees) and
increased employment. The additional employ-
ment is given by the labor demand in the farm
model; canal irrigation draws on additional ex-
ternal labor rather than household labor (a result
shown in both the double-difference and regres-
sion estimates). The distribution of this gain across
quartiles is based on the respective share of each
quartile in the incremental employment from
2005 to 2006 (table E.4). The benefit from higher
wages accrues to all those in employment and is
thus distributed across the quartiles according
to their share in 2006 employment. The whole of
the wage increase is being attributed to irriga-
tion, which is an overestimation, but these ben-
efits are relatively small, so this does not distort
the overall picture.
The allocation of net farm income is less straight-
forward, as average and marginal shares of irri-
gated land show a less clear pattern, although it
is clear that the bottom quartile has little and the
top quartile has the majority (table D.6). The cur-
rent distribution of farm income for the model is
the observed 2006 distribution, and the incre-
mental income share based on each quartile’s
share of canal and tank irrigated land in 2006.
The distribution of benefits from a hectare of ir-
rigation as a result of these calculations is shown
in figure E.1.
Community-level poverty modelTo calculate how benefits are distributed across
households at the community level, a model was
constructed using the 2006 survey data as the
baseline. The following assumptions were ap-
plied to the data:
Net farm income increases by Rs 8,800 (a weighted
average in the incremental net incomes for SRBC
and SRSP from the farm models) for each hectare
of land the household owns that will become ir-
rigated. The proportion of land irrigated is a func-
tion of income (one-fifth for the bottom quartile,
one-quarter for the second, one-third for the
third, and one-half for the top quartile).
The income of all households from agricultural
wages increased by 2.5 percent.
The gain from additional employment is calcu-
lated as a percentage increase in agricultural
SRBC SRSP
WOP WP WOP WP
Rs per hectare
Revenue 10,043 23,748 14,558 26,957
Costs 4,072 8,222 7,034 10,886
Net farm income 5,970 15,526 7,524 16,070
Wages 1,646 3,622 3,316 5,397
Days hired labor 88 188 177 280
Percentage increase
Revenue 136 85
Costs 102 55
Net farm income 160 114
Wages 120 63
Days hired labor 114 59Notes: ha = hectare; Rs = rupees; WOP = without project; WP = with project.
9 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table E.2: Estimates of the Economic Rate of Return
SAR IEG
All costs Excluding AP II ICR All costs Excluding AP II
SRBC 9 14 12 –2 0
SRSP 31 34 16 5 11
Whole project 19 24 15 2 6Note: SAR = Staff Appraisal Report.
Table E.3: Summary of Farm Model Results(average/ha)
labor income, where the increase is propor-
tionate to each quartile’s share in the marginal
increase in employment (table E.4), with an ad-
justment factor to capture the share attributable
to irrigation (the adjustment factor brings the
gains into line with those from the calculations
reported above).
The resulting income gains by quartile are shown
in figure E.2, which has a similar overall pattern
to figure E.1, the main difference being the smaller
benefit to the bottom quartile and larger benefit
to the top quartile. The model presentation breaks
down the wage and employment effects sepa-
rately. These two sets of results need not be the
same, as the former shows the distribution of in-
cremental benefits from irrigating 1 ha of land,
whereas the latter captures the effects of a com-
munity getting access to irrigation, allowing for dif-
ferential access to that irrigation.
Poverty impactThe simulation across the data set allows esti-
mation of poverty levels with and without irriga-
tion. Because the bottom quartile receives the
least benefits, the poverty impact is very sensitive
to the poverty line chosen (table E.5 and figure
E.2). The impact is also not that great because of
the finding that the bottom quartile does not
benefit from the marginal increase in employ-
ment. Relaxing that assumption and allocating
the employment benefit according to average
rather than marginal shares shows a slightly greater
poverty impact than for the lower poverty line
(scenario 2).2
The same data are used to calculate the Gini co-
efficients. These coefficients appear high because
of the large number of negative incomes (which
are set to zero for the calculation). Despite the
largest absolute benefit going to the top quartile,
the Gini is unaffected because relative income
growth is greatest among the second quartile—
or the lowest quartile, if they were to have a larger
share in the employment benefits (table E.6).
A P P E N D I X E : T H E FA R M M O D E L , R AT E O F R E T U R N A N A LY S I S , A N D D I S T R I B U T I O N A L I M PA C T S
9 5
Table E.4: Derivation of Share of Employment
Days of agriculturalemployment Share of
Quartile 2005 2006 Change change
1 278 294 16 0.06
2 123 234 111 0.40
3 159 239 80 0.29
4 62 132 70 0.25Source: IEG survey.
Figure E.1: Distribution of Benefits from 1 Hectareof Irrigation
500
01 2 3 4
Incremental wages
Incremental farm income
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
Rs p
er h
ecta
re
Quartile
Source: IEG survey.Note: Rs = rupees.
9 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Table E.5: Poverty and Inequality Measures
Upper poverty line Lower poverty line
WP WPMeasure WOP WP scenario 2 WOP WP scenario 2
Poverty headcount 0.39 0.29 0.31 0.24 0.21 0.20
Poverty gap 0.22 0.21 0.21 0.18 0.18 0.18Source: IEG survey.
Table E.6: Quartile Income Shares and Income Growth
Income shares Percentage change
Quartile Base case Simulation 1 Simulation 2 Simulation 1 Simulation 2
1 4.4 4.1 4.3 8.6 17.5
2 13.5 14.0 13.7 15.5 12.6
3 21.9 21.9 22.0 10.7 11.0
4 60.2 60.0 60.0 10.3 9.9
Figure E.2: Distribution Benefits Across All Households by Wealth Quartile
01 2 3 4
1000
2000
3000
4000
5000
6000
7000
8000
Wealth quartile
Poverty Impact of Irrigation
Inco
me
gain
per
hou
seho
ld (R
s)
Incremental wages
Incremental employment
Incremental farm income
Pove
rty
mea
sure
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
Poverty headcount
Upper poverty line
Lower poverty line
Upper poverty line
Lower poverty line
Poverty gap
WOPWP
Note: Units of measure are poverty headcount and poverty gap. Rs = rupees; WOP = base case (no irrigation); WP = with project (irrigation).
AP
PE
ND
IX E
:T
HE
FA
RM
MO
DE
L, R
AT
E O
F R
ET
UR
N A
NA
LY
SIS
, AN
D D
IST
RIB
UT
ION
AL
IMP
AC
TS
97
SRBC: Per Hectare Financial Budget: Without Project—Kharif and Two-Season Crops
Sun- Seed Vege- Ground- Cotton Sugar- Existing NewPaddy Soybean Sorghum flower bandi tables nut Cotton seeds Chilies cane Turmeric Mulberry mango mango
Yield (tonnes/ha) 3.2 0.8 1.0 0.5 0.5 7.0 1.0 1.0 0.8 0.8 80.0 1.5 4.5 4.0 8.0Price 4,000 8,000 3,500 10,750 15,000 2,200 10,500 18,500 110,000 15,000 550 15,000 8,500 5,000 5,000Value 12,800 6,400 3,500 5,375 7,500 15,400 10,500 18,500 82,500 12,000 44,000 22,500 38,250 20,000 40,000ByproductsYield (t/ha) 1.6 0.0 2.0 0.0 0.0 0.0 1.0 0.0 0.5 0.0 0.0 0.0 0.0Price 300 120 300 20,400Value 480 240 300 10,200
Total gross value 13,280 6,400 3,740 5,375 7,500 15,400 10,800 18,500 92,700 12,000 44,000 22,500 38,250 20,000 40,000InputsSeeds (kg/ha) 40 75 8 10 15 0.8 90 13 3 4 8,000 3 1,250 0 10Price (/kg) 6.9 10 5 15 30 150 20 25 200 200 0.6 1,000 0.25 0 10Value 276 750 40 150 450 120 1,800 325 600 800 4,800 3,000 3,12.5 0 100Urea (kg/ha) 100 30 60 50 10 40 100 25 40 50 75 100 300 25 150Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 400 120 240 200 40 160 400 100 160 200 300 400 1,200 100 600DAP (kg/ha) 100 30 60 50 10 40 100 25 25 50 75 100 150 35 50Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 850 255 510 425 85 340 850 212.5 425 637.5 850 1,275 297.5 425MOP (kg/ha) 0 15 0 0 0 35 0 0 0 50 35 75 150 0 150Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 75 0 0 0 175 0 0 0 250 175 375 750 0 750Manure (t/ha) 2 1 0 0 0 4 2 0 2 3 6 10 4 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140 140 140 140 140Value 280 140 0 0 0 560 280 0 280 420 840 1,400 560 0 0Chemicals (kg/ha) 2.5 0 0 0 1 4 1 0 0 5 0 1 0 1 3Price (/kg) 252 252 252 252 252 252 252 252 252 252 252 252 252 216 216Value 630 0 0 0 252 1,008 252 0 0 1,260 0 252 216 648Labor (md/ha) 170 85 53 74 96 315 84 74 800 220 364 645 346 75 85Price/md 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5Labor costs (50%) 3,188 1,594 994 1,388 1,800 5,906 1,575 1,388 15,000 4,125 6,825 12,094 6,488 1,406 1,594Price/day 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 536 402 402 369 938 436 402 670 804 1,273 804 268 67 268
Total costs 6,461 3,470 2,186 2,565 2,996 9,207 5,593 2,427 16,710 8,284 14,851 19,175 10,853 2,087 4,385Net production value 6,819 2,930 1,554 2,811 4,505 6,193 5,208 16,073 75,990 3,716 29,150 3,325 27,397 17,913 35,615
98
AN
IMP
AC
T E
VA
LU
AT
ION
OF
IND
IA’S
SE
CO
ND
AN
D T
HIR
D A
ND
HR
A P
RA
DE
SH
IRR
IGA
TIO
N P
RO
JE
CT
S
SRBC: Per Hectare Financial Budget: Without Project—Rabi Season
Rabi Sunflower SorghumSunflower Sorghum seed seeds Vegetables Groundnut Coriander Chickpea Castor Tobacco
Yield (t/ha) 0.6 1.2 0.6 1.0 7.0 1.0 0.4 0.6 0.9 1.1Price 10,750 3,500 30,000 7,500 2,200 10,500 13,000 9,350 10,000 13,000Value 6,450 4,200 18,000 7,500 15,400 10,500 5,200 5,610 9,000 14,300ByproductsYield (t/ha) 0.0 2.4 0.5 0.0 0.0 1.0 0.0 0.0 0.0 0.0Price 120 10,750 300Value 288 5,375 300
Total gross value 6,450 4,488 23,375 7,500 15,400 10,800 5,200 5,610 9,000 14,300InputsSeeds (kg/ha) 10 8 10 12 0.8 90 12 50 5 0.5Price (/kg) 15 5 45 40 150 20 15 15 45 130Value 150 40 450 480 120 1,800 180 750 225 65Urea (kg/ha) 50 60 60 120 40 100 10 30 50 100Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 200 240 240 480 160 400 40 120 200 400DAP (kg/ha) 50 50 120 150 40 100 10 30 60 120Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 425 425 1,020 1,275 340 850 85 255 510 1,020MOP (kg/ha) 0 35 0 35 35 0 0 0 0 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 175 0 175 175 0 0 0 0 0Manure (t/ha) 0 0 3 0 4 2 0 1 3 1.5Price (/t) 140 140 140 140 140 140 140 140 140 140Value 0 0 420 0 560 280 0 140 420 210Chemicals (kg/ha) 0 0 2 0 4 1 25 2 3 1Price (/kg) 252 252 252 252 252 252 4 252 252 252Value 0 0 504 0 1,008 252 100 504 756 252Labor (md/ha) 74 53 90 90 315 84 36 34 92 166Price/md 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5Labor costs (50%) 1,388 994 1,688 1,688 5,906 1,575 675 638 1,725 3,113Total animal days (adj) 12 12 14 16 28 13 10 13 18 18Price/day 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 402 402 469 536 938 436 335 436 603 603
Total costs 2,565 2,276 4,791 4,634 9,207 5,593 1,415 2,842 4,439 5,663Net production value 3,886 2,212 18,585 2,867 6,193 5,208 3,785 2,768 4,561 8,638
AP
PE
ND
IX E
:T
HE
FA
RM
MO
DE
L, R
AT
E O
F R
ET
UR
N A
NA
LY
SIS
, AN
D D
IST
RIB
UT
ION
AL
IMP
AC
TS
99
SRBC: Per Hectare Financial Budget: With Project—Kharif and Two-Season Crops
Sun- Seed Vege- Ground- Cotton Sugar- Existing NewPaddy Soybean Sorghum flower bandi tables nut Cotton seeds Chilies cane Turmeric Mulberry mango mango
Yield (t/ha) 5.4 1.5 2.5 1.6 0.8 12.0 1.6 1.5 1.2 1.6 100.0 2.0 5.0 4.5 8.0Price 4,000 8,000 3,500 10,750 15,000 2,200 10,500 18,500 110,000 15,000 550 15,000 8,500 5.000 5.000Value 21,600 12,000 8,750 17,200 12,000 26,400 16,800 27,750 132,000 24,000 55,000 30,000 42,500 22.500 40.000ByproductsYield (t/ha) 2.7 0.0 5.0 0.0 0.0 0.0 1.0 0.0 0.8 0.0 0.0 0.0 0.0Price 300 120 300 20,400Value 810 600 300 16,320
Total gross value 22,410 12,000 9,350 17,200 12,000 26,400 17,100 27,750 148,320 24,000 55,000 30,000 42,500 22.500 40.000InputsSeeds (kg/ha) 40 75 12 10 15 0.8 90 3 3 4 8,000 3 1250 0 10Price (/kg) 6.9 10 5 15 30 150 20 25 200 200 0.6 1,000 0.25 0 10Value 276 750 60 150 450 120 1,800 75 600 800 4,800 3,000 312.5 0 100Urea (kg/ha) 100 75 100 70 30 200 60 90 100 200 275 175 250 60 125Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 400 300 400 280 120 800 240 360 400 800 1,100 700 1,000 240 500DAP (kg/ha) 150 60 50 80 20 120 150 150 250 100 100 150 150 35 50Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 1,275 510 425 680 170 1,020 1,275 1275 850 850 1,275 1,275 297.5 425MOP (kg/ha) 0 35 40 0 15 100 0 200 300 100 50 100 150 0 150Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 175 200 0 75 500 0 1,000 1,500 500 250 500 750 0 750Manure (t/ha) 3 2 2 2 2 6 3 3 4 6 8 15 5 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140 140 140 140 140Value 420 280 280 280 280 840 420 420 560 840 1,120 2,100 700 0 0Chemicals (kg/ha) 3 0.5 0 1.5 2 5 1.5 5 10 7.5 0 2 0 1 3Price (/kg) 252 252 252 252 252 252 252 252 252 252 252 252 252 216 216Value 756 126 0 378 504 1260 378 1,260 2,520 1,890 0 504 216 648Labor (md/ha) 200 100 90 80 110 380 100 225 1,240 300 420 680 330 50 60Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5Labor costs (50%) 3,850 1,925 1,733 1,540 2,118 7,315 1,925 4,331 23,870 5,775 8,085 13,090 6,353 963 1,155
Total animal days (adj) 25 16 14 13 11 30 16 20 28 36 40 29 8 2 8Price/day 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 536 469 436 369 1,005 536 670 938 1,206 1,340 972 268 67 268
Total costs 7,815 4,602 3,567 3,744 4,085 12,860 6,574 9,391 30,388 12,661 17,545 22,141 10,658 1,783 3846Net production value 14,596 7,398 5,784 13,457 7,915 13,540 10,526 18,359 117,932 11,339 37,455 7,860 31,842 20,717 36,154
10
0
AN
IMP
AC
T E
VA
LU
AT
ION
OF
IND
IA’S
SE
CO
ND
AN
D T
HIR
D A
ND
HR
A P
RA
DE
SH
IRR
IGA
TIO
N P
RO
JE
CT
S
SRBC: Per Hectare Financial Budget: With Project—Rabi Season
Rabi Sunflower SorghumSunflower Sorghum seed seeds Vegetables Groundnut Coriander Chickpea Castor Tobacco
Yield (t/ha) 1.1 2.0 1.0 1.0 12.0 1.7 0.5 0.8 0.9 1.1
Price 10,750 3,500 30,000 7,500 2,200 10,500 13,000 9,350 10,000 13,000
Value 11,825 7,000 30,000 7,500 26,400 17,850 6,500 7,480 9,000 14,300Byproducts
Yield (t/ha) 0.0 4.0 0.5 0.0 0.0 1.7 0.0 0.0 0.0 0.0
Price 120 10,750 300
Value 480 5,375 510
Total gross value 11,825 7,480 35,375 7,500 26,400 18,360 6,500 7,480 9,000 143,00InputsSeeds (kg/ha) 10 12 10 12 0.8 90 20 50 5 0.5
Price (/kg) 15 5 45 40 150 20 15 15 45 130
Value 150 60 450 480 120 1,800 300 750 225 65
Urea (kg/ha) 80 90 150 120 220 100 30 40 50 100
Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0
Value 320 360 600 480 880 400 120 160 200 400
DAP (kg/ha) 100 120 120 150 25 50 10 30 60 120
Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5
Value 850 1,020 1,020 1,275 212.5 425 85 255 510 1,020
MOP (kg/ha) 0 40 50 80 100 0 0 0 0 0
Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
Value 0 200 250 400 500 0 0 0 0 0
Manure (t/ha) 2 2 3 3 6 3 0 2 3 1.5
Price (/t) 140 140 140 140 140 140 140 140 140 140
Value 280 280 420 420 840 420 0 280 420 210
Chemicals (kg/ha) 1.5 0 2 0 7 1 25 3 3 1
Price (/kg) 252 252 252 252 252 252 4 252 252 252
Value 378 0 504 0 1,764 252 100 756 756 252
Labor (md/ha) 90 70 90 80 380 105 40 40 92 166
Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5
Labor costs (50%) 1,733 1,348 1,733 1,540 7,315 2,021 770 770 1,771 3,196
Total animal days (adj) 13 14 14 16 34 18 10 13 18 18
Price/day 50 50 50 50 50 50 50 50 50 50
Animal costs (67%) 436 469 469 536 1139 603 335 436 603 603
Total costs 4,146 3,737 5,446 5,131 12,771 5,921 1,710 3,407 4,485 5,746
Net production value 7,679 3,744 29,930 2,369 13,630 12,439 4,790 4,074 4,515 8,555
AP
PE
ND
IX E
:T
HE
FA
RM
MO
DE
L, R
AT
E O
F R
ET
UR
N A
NA
LY
SIS
, AN
D D
IST
RIB
UT
ION
AL
IMP
AC
TS
10
1
SRBC: Per Hectare Financial Budget: Without Project
Kharif crops Two season Rabi crops
ExistingPaddy Maize Pulses Groundnut Cotton Chilies Turmeric mango Sunflower Paddy Maize Groundnut Pulses
Yield (t/ha) 3.0 2.0 0.6 0.6 0.8 0.8 2.5 4.0 0.6 3.0 2.2 1.0 0.6Price 4,000 3,800 10,500 10,500 18,500 15,000 15,000 5,000 10,750 4,000 3,800 10,500 10,500Value 12,000 7,600 6,300 6,300 14,800 12,000 37,500 20,000 6,450 12,000 8,360 10,500 6,300ByproductsYield (t/ha) 1.5 2.0 0.0 1.0 0.0 0.0 0.0 0.0 0.0 1.5 2.2 1.0 0.0Price 300 80 300 300 80 300Value 450 160 300 450 176 300
Total gross value 12,450 7,760 6,300 6,600 14,800 12,000 37,500 20,000 6,450 12,450 8,536 10,800 6,300InputsSeeds (kg/ha) 40 20 20 90 13 4 3 0 10 40 20 90 20Price (/kg) 6.9 10 10 20 25 200 1000 0 15 6.9 10 20 10Value 276 200 200 1800 325 800 3000 0 150 276 200 1,800 200Urea (kg/ha) 150 120 10 40 30 100 150 25 50 150 120 40 30Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 600 480 40 160 120 400 600 100 200 600 480 160 120DAP (kg/ha) 100 50 50 100 25 50 100 35 50 100 50 100 50Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 850 425 425 850 212.5 425 850 297.5 425 850 425 850 425MOP (kg/ha) 0 30 0 0 0 50 75 0 0 0 30 0 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 150 0 0 0 250 375 0 0 0 150 0 0Manure (t/ha) 2 3 0 2 0 3 10 0 0 2 3 2 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140 140 140Value 280 420 0 280 0 420 1400 0 0 280 420 280 0Chemicals (kg/ha) 2.5 2 0 1 0 5 1 1 0 2.5 2 1 0Price (/kg) 252 252 252 252 252 252 252 216 252 252 252 252 252Value 630 504 0 252 0 1260 252 216 0 630 504 252 0Labor (md/ha) 170 98 78 103 84 245 645 75 78 180 98 103 57Price/md 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5Labor costs (50%) 3,188 1,838 1,463 1,931 1,575 4,594 12,094 1,406 1,463 3,375 1,838 1,931 1,069
Total animal days (adj) 25 18 10 13 12 24 24 2 12 25 18 13 10Price/day 50 50 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 603 335 436 402 804 804 67 402 838 603 436 335
Total costs 6,661 4,620 2,463 5,709 2,635 8,953 19,375 2,087 2,640 6,849 4,620 5,709 2,149Net production value 5,789 3,141 3,838 891 12,166 3,047 18,125 17,913 3,811 5,602 3,917 5,091 4,151
10
2
AN
IMP
AC
T E
VA
LU
AT
ION
OF
IND
IA’S
SE
CO
ND
AN
D T
HIR
D A
ND
HR
A P
RA
DE
SH
IRR
IGA
TIO
N P
RO
JE
CT
S
SRBC: Per Hectare Financial Budget: With Project—Kharif and Two-Season Crops
ExistingPaddy Maize Sunflower Pulses Vegetables Groundnut Cotton Chilies Turmeric mango Soybean
Yield (t/ha) 4.2 2.2 1 0.9 12.0 1.2 1.6 1.6 3.5 4.5 0.5Price 4,000 3,800 10,750 10,500 2,200 10,500 18,500 15,000 15,000 5,000 8,000Value 16,800 8,360 10,750 9,450 26,400 12,600 29,600 24,000 52,500 22,500 4,000ByproductsYield (t/ha) 2.1 2.2 0.0 1.0 0.0 0.0 0.0 0.0Price 300 80 300Value 630 176 300
Total gross value 17,430 8536 10,750 9,450 26,400 12,900 29,600 24,000 52,500 22,500 4,000InputsSeeds (kg/ha) 40 20 10 20 0.8 90 3 4 3 0 75Price (/kg) 6.9 20 15 10 150 20 25 200 1,000 0 10Value 276 400 150 200 120 1,800 75 800 3,000 0 750Urea (kg/ha) 200 160 50 35 150 70 90 200 175 35 75Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 800 640 200 140 600 280 360 800 700 140 300DAP (kg/ha) 125 65 70 65 100 125 100 75 125 40 75Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 1,062.5 552.5 595 552.5 850 1,062.5 850 637.5 1,062.5 340 637.5MOP (kg/ha) 0 60 60 0 70 0 0 75 90 0 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 300 300 0 350 0 0 375 450 0 0Manure (t/ha) 2 3 3 0 0 2 0 3 10 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140Value 280 420 420 0 0 280 0 420 1,400 0 0Chemicals (kg/ha) 3.5 3 1 0.5 4 1 5 7 1.5 1.5 0.5Price (/kg) 252 252 252 252 252 252 252 252 252 216 252Value 882 756 252 126 1,008 252 1,260 1764 378 324 126Labor (md/ha) 205 110 88 65 300 110 180 245 710 75 80Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5Labor costs (50%) 3,946 2,118 1,694 1,251 5,775 2,118 3,465 4,716 13,668 1,444 1,540
Total animal days (adj) 25 18 16 10 25 15 18 30 26 2 10Price/day 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 603 536 335 838 503 603 1,005 871 67 335
Total costs 8,084 5,789 4,147 2,605 9,541 6,295 6,613 10,518 21,529 2,315 3,689Net production value 9,346 2,747 6,603 6,845 16,860 6,606 22,987 13,482 30,971 20,185 312
AP
PE
ND
IX E
:T
HE
FA
RM
MO
DE
L, R
AT
E O
F R
ET
UR
N A
NA
LY
SIS
, AN
D D
IST
RIB
UT
ION
AL
IMP
AC
TS
10
3
SRBC: Per Hectare Financial Budget: With Project—Rabi Season
ExistingChilies Turmeric mango Sunflower Paddy Maize Vegetables Groundnut Pulses Soybean
Yield (t/ha) 1.6 3.5 4.5 1.5 4.2 3.2 12.0 1.8 1.0 0.5Price 15,000 15,000 5,000 10,750 4,000 3,800 2,200 10,500 10,500 8,000Value 24,000 52,500 22,500 16,125 16,800 12,160 26,400 18,900 10,500 4,000ByproductsYield (t/ha) 0.0 0.0 0.0 0.0 2.1 3.2 1.8 0.0Price 300 80 300Value 630 256 540
Total gross value 24,000 52,500 22,500 16,125 17,430 12,416 26,400 19,440 10,500 4,000InputsSeeds (kg/ha) 4 3 0 10 40 20 0.8 90 20 75Price (/kg) 200 1,000 0 15 6.9 10 150 20 10 10Value 800 3,000 0 150 276 200 120 1800 200 750Urea (kg/ha) 200 175 35 50 200 60 150 70 35 75Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 800 700 140 200 800 240 600 280 140 300DAP (kg/ha) 75 125 40 50 125 65 100 125 65 75Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 637.5 1,062.5 340 425 1,062.5 552.5 850 1,062.5 552.5 637.5MOP (kg/ha) 75 90 0 0 0 60 70 0 20 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 375 450 0 0 0 300 350 0 100 0Manure (t/ha) 3 10 0 0 2 3 3 2 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140Value 420 1,400 0 0 280 420 420 280 0 0Chemicals (kg/ha) 7 1.5 1.5 1 3.5 3 4 1 0.5 0.5Price (/kg) 252 252 216 252 252 252 252 252 252 252Value 1,764 378 324 252 882 756 1,008 252 126 126Labor (md/ha) 245 710 75 90 205 122 250 110 65 80Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5Labor costs (50%) 4,716 13,668 1,444 1,733 3,946 2,349 4,813 2,118 1,251 1,540
Total animal days (adj) 30 26 2 13 18 20 25 16 16 10Price/day 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 1,005 871 67 436 603 670 838 536 536 335
Total costs 10,518 21,529 2,315 3,195 7,850 5,487 8,998 6,328 2,906 3,689Net production value 13,482 30,971 20,185 12,930 9,580 6,929 17,402 13,112 7,594 312
1 0 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
SRBC Economic Rate of Return (Rs)
Investment Addtitional Total Incremental NetYear costs O&M costs benefit benefit
1989 284 0 284 0 –284
1990 347 0 347 0 –347
1991 294 0 294 0 –294
1992 1,005 0 1,005 0 –1,005
1993 918 0 918 0 –918
1994 2,354 0 2,354 0 –2,354
1995 2,843 0 2,843 0 –2,843
1996 768 0 768 0 –768
1997 787 0 787 0 –787
1998 578 0 578 0 –578
1999 714 0 714 0 –714
2000 1,148 0 1,148 0 –1,148
2001 1,287 0 1,287 0 –1,287
2002 1,348 0 1,348 0 –1,348
2003 867 0 867 0 –867
2004 827 5 832 74 –758
2005 0 5 5 74 69
2006 0 16 16 263 247
2007 0 23 23 368 346
2008 0 33 33 547 514
2009 0 33 33 547 514
2010 0 33 33 547 514
2011 0 33 33 547 514
2012 0 33 33 547 514
2013 0 33 33 547 514
2014 0 33 33 547 514
2015 0 33 33 547 514
2016 0 33 33 547 514
2017 0 33 33 547 514
2018 0 33 33 547 514
2019 0 33 33 547 514
2020 0 33 33 547 514
2021 0 33 33 547 514
2022 0 33 33 547 514
2023 0 33 33 547 514
2024 0 33 33 547 514
2025 0 33 33 547 514
2026 0 33 33 547 514
2027 0 33 33 547 514
2028 0 33 33 547 514Note: ERR = –2 percent. Rs = rupees.
A P P E N D I X E : T H E FA R M M O D E L , R AT E O F R E T U R N A N A LY S I S , A N D D I S T R I B U T I O N A L I M PA C T S
1 0 5
SRSP Economic Rate of Return (Rs)
Investment Addtitional Total Incremental NetYear costs O&M costs benefit benefit
1989 3,581 0 3,581 0 –3,581
1990 431 0 431 0 –431
1991 380 0 380 0 –380
1992 633 0 633 0 –633
1993 694 0 694 0 –694
1994 890 0 890 0 –890
1995 1,245 0 1,245 0 –1,245
1996 728 16 744 216 –528
1997 1,643 16 1,660 216 –1,443
1998 1,209 49 1,257 649 –609
1999 1,456 65 1,522 865 –657
2000 1,863 81 1,944 1,081 –863
2001 2,623 90 2,713 1,189 –1,524
2002 1,156 98 1,254 1,297 43
2003 2,027 106 2,133 1,405 –728
2004 1,549 114 1,664 1,514 –150
2005 0 122 122 1,622 1,499
2006 0 130 130 1,730 1,599
2007 0 130 130 1,730 1,599
2008 0 130 130 1,730 1,599
2009 0 130 130 1,730 1,599
2010 0 130 130 1,730 1,599
2011 0 130 130 1,730 1,599
2012 0 130 130 1,730 1,599
2013 0 130 130 1,730 1,599
2014 0 130 130 1,730 1,599
2015 0 130 130 1,730 1,599
2016 0 130 130 1,730 1,599
2017 0 130 130 1,730 1,599
2018 0 130 130 1,730 1,599
2019 0 130 130 1,730 1,599
2020 0 130 130 1,730 1,599
2021 0 130 130 1,730 1,599
2022 0 130 130 1,730 1,599
2023 0 130 130 1,730 1,599
2024 0 130 130 1,730 1,599
2025 0 130 130 1,730 1,599
2026 0 130 130 1,730 1,599
2027 0 130 130 1,730 1,599
2028 0 130 130 1,730 1,599Note: ERR = 5 percent. Rs = rupees.
1 0 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Combined SRBC and SRSP Economic Rate of Return (Rs)
Investment Addtitional Total Incremental NetYear costs O&M costs benefit benefit
1989 3,865 0 3,865 0 –3,865
1990 778 0 7,78 0 –778
1991 675 0 675 0 –675
1992 1,639 0 1,639 0 –1,639
1993 1,612 0 1,612 0 –1,612
1994 3,245 0 3,245 0 –3,245
1995 4,088 0 4,088 0 –4,088
1996 1,496 16 1,512 216 –1,296
1997 2,430 16 2,447 216 –2,230
1998 1,786 49 1,835 649 –1,187
1999 2,170 65 2,236 865 –1,371
2000 3,011 81 3,092 1,081 –2,011
2001 3,910 90 4,000 1,189 –2,811
2002 2,504 98 2,602 1,297 –1,304
2003 2,895 106 3,001 1,405 –1,595
2004 2,377 119 2,495 1,587 –908
2005 0 127 127 1,695 1,569
2006 0 146 146 1,993 1,847
2007 0 153 153 2,098 1,945
2008 0 164 164 2,277 2,113
2009 0 164 164 2,277 2,113
2010 0 164 164 2,277 2,113
2011 0 164 164 2,277 2,113
2012 0 164 164 2,277 2,113
2013 0 164 164 2,277 2,113
2014 0 164 164 2,277 2,113
2015 0 164 164 2,277 2,113
2016 0 164 164 2,277 2,113
2017 0 164 164 2,277 2,113
2018 0 164 164 2,277 2,113
2019 0 164 164 2,277 2,113
2020 0 164 164 2,277 2,113
2021 0 164 164 2,277 2,113
2022 0 164 164 2,277 2,113
2023 0 164 164 2,277 2,113
2024 0 164 164 2,277 2,113
2025 0 164 164 2,277 2,113
2026 0 164 164 2,277 2,113
2027 0 164 164 2,277 2,113
2028 0 164 164 2,277 2,113
Note: ERR = 2 percent. Costs are investment costs, excluding dam safety, O&M, and land costs. All prices converted to 2006 prices, using wholesale priceindex deflator where necessary. Incremental benefit is net farm income from farm model (valued at economic prices) times newly irrigated area. Standardconversion factor is 0.92. Rs = rupees.
10
8
AN
IMP
AC
T E
VA
LU
AT
ION
OF
IND
IA’S
SE
CO
ND
AN
D T
HIR
D A
ND
HR
A P
RA
DE
SH
IRR
IGA
TIO
N P
RO
JE
CT
S
Appendix F: Village Profiles
WUA member(% all Irrigation source (% of all plots)
households) 2005 2006
Village Water availability Canal condition Reach 2005 2006 Canal Tank Canal Tank
Uyyalawada No water Canal not completed because Tail 0 2 1.6 0.0 0.0 1.5of land dispute; heavy siltation
Vennaram Received water once only in 2004 Canal not completed Tail 12 4 4.1 0.0 4.0 0.0
Lingagiri No water Not yet constructed n.a. 2 0 0.0 9.6 0.0 23.6
Pragthisingaram No water Not yet constructed n.a. 2 14 0.0 10.3 5.0 62.5
Narsapur Water since rabi 2006 n.a. n.a. 2 20 0.0 11.4 13.6 44.1
Ingurthy Water since 2005; most goes to tanks Good Head 10 14 17.1 2.4 11.1 8.9
Komatipalli Water since 2005, goes to tanks preventing Main distributary Middle 18 10 2.9 1.4 7.0 8.5flow to two villages downstream clogged, tertiary good
Upparagudem Received water once only in 2004 Canal completed by Middle 0 12 0.0 18.9 2.6 15.4heavily silted
Kampalle Received water once only in 2004. Heavily silted Tail 0 8 0.0 0.0 0.0 13.0Water taken by farmers upstream
Nallabelli Water since rabi 2006 n.a. n.a. 10 28 2.4 14.3 4.0 44.0
Rajanpalli No water Not yet constructed n.a. 12 18 5.6 9.3 0.0 12.7
Malaya Water since 2003, flows to tanks Main distributary good, Middle 0 4 6.3 4.2 5.6 9.3tertiaries clogged
Mahadesapuram Received water once only in 2004 Heavily silted Head 0 0 0.0 2.0 0.0 1.9
Thopanpally Water since 2003, goes to tanks preventing Good Head 12 14 11.9 2.4 17.8 6.7flow to three villages downstream
Deekshakunta Water since 2005, flows to tank Good Middle 13 19 2.2 13.0 3.7 22.2
Peddakorpolu No water, diverted upstream Distributary unusable Tail 10 0 0.0 11.5 0.0 12.3to other villages
Chinnakondapak Water not released because farmers n.a. n.a. 0 20 0.0 26.8 0.0 30.4want to release to tanks to which ID opposed
Muchimpala Canal Good Middle 10 12 0.0 8.6 12.8 25.6Note: n.a. = not applicable.
1 0 9
I. Introduction1
Irrigation is definitely a powerful force for poverty
reduction. The expansion in the irrigation sector
and adoption of new techniques of agriculture
(Green Revolution) has resulted in an increase of
per capita grain production, which clearly shows
that the reason behind it is irrigation. Hence, ir-
rigation cannot be concluded as a paradox, and
it is the basic requirement for the sustenance of
mankind. Any developmental activity has got its
diversified impact, and a balanced view has to be
taken when an assessment is being taken. In-
vestments made toward irrigation infrastructure
are mainly one time, but the benefits (direct/
indirect) are long term. The emphasis on new
construction, rehabilitation of systems, and qual-
ity of work is altogether different from the
schemes planned especially to eradicate poverty.
It is agreed that the findings of the IEG study are
important and definitely provide altogether dif-
ferent perceptions, but to draw conclusions on the
entire irrigation development portfolio is in-
appropriate and not justifiable. The irrigation in-
frastructure creation as an effective mechanism for
fostering growth and economic upliftment in the
project areas is proved beyond doubt across the
globe. Moreover, the methodology of assessing the
impact, the sample size, and the sample chosen
by IEG has significant differences, and therefore,
the findings are misleading.
It is very much understandable that the two proj-
ects taken up are of different nature; that is, one
is basically a rehabilitation of an existing system
(SRSP) and the other one is set to complete a new
irrigation system for bringing new areas under
irrigation. Thus, special issues concerned with
these two projects and circumstances considered
by the Bank for assisting are overlooked by the
study, and the report is made exclusive of these
facts. It may not be out of context to mention that
IEG has prejudged the issues and extrapolated/
generalized the small sample survey results across
the irrigation investments.
However, the project-specific status with re-
marks and participatory irrigation management
efforts taken up by the state are highlighted
below so that the objective assessment of the
AP II and AP III projects is brought out in a more
focused manner.
II. Performance of SRBC and SRSPProjects
SRBC project Salient features of SRBC and the present progress
of the project:
• The SRBC scheme was formulated to irrigate
190,000 acres by utilizing 19 TMC of Krishna
water from the foreshore of Srisailam Dam to
benefit the chronic drought-prone areas in
Nandyal, Banaganapalli, and Koilakuntla taluks
of Kurnool district and Jammalamadugu taluk
of Kadapa district.
• The SRBC scheme was cleared by the plan-
ning commission in 1981.
• After completion of preliminaries, the works
from km 0–km 10 of SRBC were commenced
in 1984.
• Because of paucity of funds and to aid in the
speedy execution of the project, part of the
scheme was posed for World Bank assistance
in 1984.
APPENDIX G: COMMENTS FROM GOVERNMENT OF ANDHRA PRADESH
• 37 percent of the works were completed by the
time AP II was closed in 1994.
• The balance of the works of the project (except
Gorakallu balancing Reservoir, Owk second
stage, and main canal beyond 141.00 to 199.00
under SRBC scheme) was taken up with World
Bank assistance under a separate project—
AP III.
• Under AP II and AP III, all infrastructure to cre-
ate irrigation potential to an extent of 1,53,936
acres under 16 blocks was completed except
a micro-network.
Status of the Project: As of March 2004, Rs.
988.52 Crores had been spent, and the main canal
and distribution system were completed; as of
December 2006, 1,192.46 Crores had been spent
and water supplied through field channels to an
extent of 88,121 acres.
Works in Progress in SRBC: Sri Narasimharaya
Sagar Project—
1. Excavation of field channels:
(a) From Block I to X by Ayacutdar Committees/
water user associations (WUAs)
(b) From Block XI to XIV by Contracting agency
(c) Phase II works of Owk Reservoir Complex.
2. SRBC from km 141 to km 199, including dis-
tributaries XVII to XXI and micro-network.
Present Scenario: Expenditure particulars under
World Bank aid
Andhra Pradesh II irrigation project—333
Crores (approximately)
AP III irrigation project—845 Crores (approx-
imately)
(a) Micro-network under XV and XVI blocks
were completed.
(b) Micro-network under I to X blocks are
being executed under CADWM program.
(c) Micro-network under XI to XIV are being
executed by contracting agency.
(d) Irrigation potential created up to 2006
under I to XVI blocks is 153,936 acres and
ayacut brought under irrigation is 87,895
acres by constructing field channels.
(e) Since 2004, until water is supplied through
the SRBC system to an ayacut of 30,000 A/c
to 130,000 A/c as per the ayacut develop-
ment, as shown below (includes without
field channels also):
2004–30,000 A/c
2005–50,000 A/c against 160,000 A/c
2006–130,000 A/c }(f) Hence the benefits are now reaching the
beneficiaries. Moreover, the WUAs were re-
cently elected and have been functioning
since November 2006.
(g) The sample survey for SRBC will reveal a
different picture if taken up now as it is pro-
grammed to release water for entire aya-
cut of 160,000 A/c during kharif 2007.
(h) Furthermore, the CWC is conducting
monitoring every year and communicat-
ing its observations and the same are
being complied.
With regard to the sample selection of IEG irri-
gation survey, it has considered two command
areas in SRBC and SRSP. At the time of the first
round of the survey, very little water had been re-
leased, irrigating 7,800 ha of a planned 65,000 ha
in the SRBC command, and the simple reason
is that the micro-network was completed to an
extent of 20 percent. Completion of the network
was estimated for 2008, so IEG has not undertaken
the second round of a survey in 2006 in the
SRBC command. This is to say that 12 percent
of lands were able to be irrigated, and if no
system is developed, it could have been simply nil.
The deficiency of 8 percent cannot be directly
called a shortfall, as the farmers will take sufficient
time to develop their lands even after water is
made available to them, due to their financial
and other constraints. This is quite evident
when we observe the development of ayacut
in most successful schemes like Godavari Delta,
Krishna Delta, and the Nagarjuna Sagar Project.
Further, SRBC was a new scheme, whereas SRSP
was an extension of existing system under
AP II.
1 1 0
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
The government departments will be strength-
ened in due time for maintaining efficient man-
agement of the irrigation system, so as to redress
the weakness in the activities of WUAs.
IEG has come to a conclusion that the project’s fail-
ure can be attributed to problems both in design
and implementation. The rates that were adopted
in the estimates are purely based on realistic val-
ues and hence only the contractors are able to
execute the works with discounts. Further, con-
cluding that government officials at all levels ben-
efited financially from irrigation source, from the
study of Robert Wade (1982) in itself is not ac-
ceptable to the state and the same cannot be sub-
stantiated and the statement is not in good taste.
Sriramasagar projectSalient features of the project: The Srirama-
sagar project was constructed across the River
Godavari to create an irrigation potential of
968,640 acres in four districts: viz., Adilabad,
Nizamabad, Karimnagar, and Warangal. Before
taking up AP III, 630,000 acres of irrigation po-
tential was created. The area irrigated during
1995–96 was 268,301 acres, against the irrigation
potential creation of 630,000 acres.
The third Andhra Pradesh irrigation potential was
taken up to complete the irrigation development
and scheme rehabilitation works began under
second Andhra Pradesh irrigation potential and
thus realize the potential for increasing agricultural
production and rural income in backward regions
of Telangana, covering the districts of Adilabad,
Nizamabad, Karimnagar, and Warangal. Rehabili-
tation and modernization of Kakatiya Canal from
0 km to 234 km and distributory system to stabi-
lize the aycut of 625,000 acres.
District breakdown:
Nizamabad: 11,000 acresKarimnagar: 543,000 acresWarangal: 71,000 acres
Total: 625,000 acres
Objectives of AP III:
• Stabilization of 625,000 acres ayacut of Kakatiya
Canal up to 234 km
• Dam safety works (SRS Dam, Lower Manair
Dam) as recommended by Dam Safety Panel
• Environment Management Plan
• Agricultural support services
• R&R of project-affected persons
• O&M works.
Achievements:
The activities envisaged under AP III are imple-
mented. As a result the carrying capacity of
Kakatiya Canal increased from 8,250 cubic feet per
second (cusecs) to 9,700 cusecs. Thus the water
reached to tail-end areas of all the distributaries,
that is, from 0 to 234 km. The area irrigated also
increased from 268,301 acres to 796,887 acres
(2001–2002).
The following comments are offered on the Re-
port prepared by IEG on impact evaluation of AP
II and AP III projects in respect of Sriramasagar
Project:
• AP II could not be completed as planned mainly
because of delay in finalization of an R&R plan.
There was considerable delay in finalizing
model bid documents for civil works. The World
Bank indicated that it preferred to close AP II
instead of extending the project by 2–3 years.
• Appraisal for AP III, which commenced in Feb-
ruary 1994 (just prior to closure of AP II in June
1994), took three years.
• AP III became effective in July 1997. The clos-
ing date was extended by 18 months and the
project was closed in June 2004.
• Even though some problems in procurement
were encountered initially, AP III closed on a
successful note. IEG has made drastic com-
ments on the functioning of government de-
partments. It is felt desirable not to make an
issue of these sorts of comments.
• However, to put the issues in proper perspec-
tive the following points are deemed necessary.
The issue of changing to crops other than
paddy defies solution, as seen from past ex-
perience. It is better if this issue is not high-
lighted any longer. During the course of AP
III appraisal, it was hinted that supplies from
the canal should be restricted to ID crop
requirement. The farmers can supplement
A P P E N D I X G : C O M M E N T S F R O M G O V E R N M E N T O F A N D H R A P R A D E S H
1 1 1
balance requirement from ground water
or restrict their irrigated area.
Without having a first-hand knowl-
edge of problems encountered in im-
plementation of AP II and AP III, making
light of the work done achievements is
not fair.
Several supervision missions of Bank
study teams have visited the project dur-
ing implementation and appreciated the
work being done. The manner in which
the benefits of AP III are assessed and
quantified may differ from agency to
agency.
Verifiable benefits of civil works under AP III:
(a) Enhanced carrying capacity of Kakatiya Canal
from 8,250 cusecs to 9,700 cusecs. Records
maintained in the department will prove this
point.
(b) The distribution system that had deterio-
rated over a period of time was put back in
place. The area irrigated increased over a pe-
riod of time, with completion of rehabilitation
and modernization works. Reports of various
supervision missions contain the details.
(i) Before AP III, the maximum discharge in
to the main canal was to 6,000 cusecs ver-
sus 8,250 cusecs.
(ii) After taking up and completion of AP
III, the canal discharge was greater than
9,000 cusecs.
(c) The achievements listed in (a) and (b) above
were possible with the same quantum of
water. The inflows in to the reservoirs and re-
leases over several years can be verified.
(d) Productivity has considerably increased and
so has the income of farmers.
(e) Farmers’ organizations, which were put in place
in 1997 for a 5-year term, functioned fairly well.
However, it has to be put on record that the con-
clusions of the study do not adequately represent
ground reality. The conclusions are not based on
project-specific data.
III. General Remarks about Sample Size The survey consisted of two rounds in June 2005
and March 2006, with limited interrogation of 50
households in each village of the 16 total villages
covered and the data collected on income, farm
budgets, irrigation, and various aspects of social
and political life in the village cannot be formed
as a base, especially in the countries like India,
which is in developing stages. Thus, the conclu-
sions formulated based on the limited survey
with inadvertent public cannot form a base to
extrapolate the entire situation and thus the con-
clusions will not be realistic.
The study, which is a sample one, has revealed cer-
tain things. A sample can be called as a viable
sample if it satisfies the requirements of certain
principles. Further, conclusions made on SRBC
based on sample of SRSP are inappropriate.
IV. Participatory Irrigation ManagementTo improve irrigation performance, the govern-
ment of Andhra Pradesh took a progressive and
innovative step to empower the farmers to man-
age and operate the irrigation resources through
formation of WUAs in 1997. The exclusive Andhra
Pradesh Farmers Managed Irrigation Act of 1997
was enacted, which provides the legal support for
the functioning of these WUAs.
The objectives of farmers’ organizations are car-
rying O&M of the irrigation system, effective water
management, and increased agricultural pro-
duction. The act emphasized formation of water
users’ organizations at three levels, namely WUAs,
distributory committees (DCs) and project com-
mittees for major irrigation projects; two-tier
structure (WUA/PC) for medium irrigation proj-
ects; and single-tier structures (WUA) for minor
irrigation.
The following remarks are made on IEG’s obser-
vations regarding participatory irrigation man-
agement (PIM).
1. The expansion and durability of WUAs
has been influenced by external forces.
In 1997, WUAs were constituted for all the irri-
gation sectors and numbered around 10,000 for
involvement of the stakeholders in the man-
agement of irrigation systems. DCs were also
constituted.
1 1 2
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
After completion of the WUAs’ 5-year tenure, the
performance of these bodies is evaluated and
thought to bring changes in the set-up of the
WUAs for more accountability and more con-
centration on other statutory requirements, in-
cluding water management.
It is felt that the WUAs are to be made continu-
ous bodies to have a blend of old and new mem-
bers for better working of these organizations.
To bring the above changes, there was a gap of
about one year for conduct of elections to the
WUAs.
Elections to all the WUAs in the state were com-
pleted and elections for the DCs were also com-
pleted in 13 districts and will be completed by the
end of April 2007 in the remaining districts. WUAs
and DCs are also constituted in the SRBC project
during November 2006. Constitution of project
committees is also under active consideration of
the government.
2. WUA membership is not representative
of the population served.
IEG observed that half of the households in 18
villages of Warangal district have access to canal
or tank irrigation, but only 18 percent of the all
households are members. In this connection, the
Act provided membership only for the land hold-
ers; that is, land is to be in the name of the per-
son who has voting rights and who is a member
of the association. The tenant will get voting
rights and be a member of the association if the
land holder leases the land. Normally, the land
is registered in the name of the head of the fam-
ily; thus, the households are represented in a
smaller percentage.
Representation of the WUAs is mostly by the elite
groups; the study conducted by the IEG group is
only for a limited area and for all practical purposes
the study conducted by ORG (2005) can be taken
as reliable study, as the sample taken for the same
is 214 WUAs across the state and in all three sec-
tors of irrigation. Backward castes represented
around 60 percent of the managing committee of
the WUAs. The study also revealed that around 50
percent of the marginal farmers are represented
in the managing committee of the WUAs.
Thus, it is not correct that most of the managing
committees of WUAs are represented by elite
groups. In this connection, it is informed that all
the land holders are the voters of the WUA and
these are modeled as self-help groups; there are
no reservations based on social status. However,
in the future, the representation of small and
marginal farmers and also other social groups
will also increase over and above the present
level, as is happening in the other institutions.
Regarding benefits to the poor, the AP II and AP
III projects are not exclusively meant for poverty
eradication, and in case of irrigation projects, in-
tangible benefits will be more and the same are
not emphasized in the IEG study.
A gap ayacut of 10.07 lakhs is bridged during the
first three years of implementation of PIM. This
will definitely help the poor bring the livelihood
opportunities by way of increasing the number of
man working days.
3. A positive feature of irrigation devel-
opment since the 1997 reforms has been
rehabilitation of sections of the canal
network, so that water now reaches more
areas.
After formation of the WUAs and DCs in 1997,
massive O&M and MR works were taken in all
commands and farmers decided the works should
be taken up; they are actively involved in the ex-
ecution. However, there may be some defects in
certain cases; for example, WUA presidents did
not take other members into confidence and
benefited personally as contractors. This was cor-
rected, making the president’s election indirect,
which necessarily makes the decisions collec-
tive. WUAs are made continuous bodies. The
term of 5 years has resulted in complacency, so
the term of the president is 2 years to address this
problem.
The water rates could not be collected to the ex-
pected level due to continuous drought. It can be
observed that the water tax collections are around
A P P E N D I X G : C O M M E N T S F R O M G O V E R N M E N T O F A N D H R A P R A D E S H
1 1 3
100 Crores during 1998–99, 1999–2000, and
2000–01 as the season was good. Now O&M is
linked to the water tax collections, and the farm-
ers are being informed time and again that more
collections will bring sufficient funds for O&M.
It is not correct that the reforms have not broken
a vicious circle, as the study is limited to the SRSP
Project only, and there are no WUAs in SRBC;
thus, the study may not reflect the true picture of
PIM. The benefits of the PIM are:
• Sense of ownership
• Dynamic leadership
• Execution of works by WUA
• 10.07 lakh acres gap ayacut bridge
• Increased agricultural production.
As per the ORG study on the performance of the
WUAs, around 70 percent WUAs performance is
assessed above average in major and minor sec-
tors and 60 percent in medium sector [IEG note:
these figures are, in fact 50, 51, and 51 percent,
respectively; ORG 2005].
4. The farmers’ committee structure (WUAs,
DCs, and project committees) is intended
to ensure equitable allocation of water, but
the structure is incomplete and has lim-
ited capacity to do this.
During 1997 elections to the WUAs and DCs were
conducted, and they have taken over the charge
of O&M and water management. However, proj-
ect committees could not be formed, as it was felt
that the office bearers of the WUAs and DCs get
well experienced with the management of the ir-
rigation systems before they take up the re-
sponsibility of the project committee, which
requires better administrative and management
skills.
It is mentioned in the report that breaches of the
canal to take water are also common and WUAs
seem powerless to act. Indeed, they can act as ve-
hicles for elite capture of these resources. The
ORG study (2005) revealed that around 60 per-
cent of the farmers believe that the water supply
is more reliable and supply has increased.
Another study conducted by Dr. K.V. Raju (2001)
reports that in 50 percent of tail-end WUAs, farm-
ers felt the water has reached their fields with the
activities taken up by the WUAs. It is also reported
that there is considerable reduction in water
conflicts.
As these institutions are new, it cannot be ex-
pected that results will be at 100 percent. In the
first five years the WUAs and DCs actively involved
in the system maintenance. They have also par-
ticipated in water management. They have pri-
oritized and executed the work. The next step is
toward better water management. In this direc-
tion steps are being taken by the government
through training and capacity building.
5. Are WUAs an effective means to provide
sustainable O&M?
The O&M is linked to the water tax collections and
FOs. In the major and medium sector, FOs will get
50 percent, and in the minor sector 90 percent of
the water tax collections will go toward O&M. In
the previous year an amount of Rs 35 Crores was
released toward O&M and this year an amount of
Rs 60 Crores is earmarked for O&M. This will go
up to Rs 150 Crores in the coming years, and
O&M works can be taken up in all commands to
satisfactory level.
6. Bottom-up development needs support
from the top.
The government of Andhra Pradesh is keen to im-
plement PIM in the irrigation sector. As such,
elections to all the WUAs have been completed
and DCs are also constituted in certain projects;
in the remaining projects it will be completed by
the end of April 2007. As such, it is not correct that
there is a lack of support for WUAs, and govern-
ment is committed for stakeholders’ participa-
tion in management of irrigation systems.
7. WUAs have only limited means to re-
solve water-allocation disputes.
It is not correct to say that WUAs have only lim-
ited means to resolve water-allocation disputes per
the provisions of the Act. The farmers’ organiza-
tions have been given powers for settlement of dis-
putes (§§ 26, 27 of the Act).
1 1 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
V. Conclusions• Based on limited study, conclusions, if any are
drawn, will not be considered realistic.
• Visualizing the estimates either low or high is
probable in the sense that the features on the
ground cannot be foreseen accurately.
• Educating the farmers in a scientific way takes
much time. Processes are taking time.
• Farmers will be satisfied with the scientific
approaches only if they realize them on the
field.
• The success will depend on the availability of
water and the timing.
• The output/productivity largely depends on
many factors, like monsoons, pest control,
other management practices, and so forth.
• The farmers should get loans for their inputs, such
as for seeds, pesticides, fertilizers, and so forth.
• Only through sustained campaign and capac-
ity building activities can 100 percent water ac-
cess can be realized, leading to sustenance of
the systems.
• The reform process will continue in the irri-
gation sector, and the stakeholders will be
involved more in coming years for better man-
agement of the systems.
A P P E N D I X G : C O M M E N T S F R O M G O V E R N M E N T O F A N D H R A P R A D E S H
1 1 5
1 1 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
Attachment 1: Details of Ayacut Contemplated and Irrigated under SRBC in I to XVI Blocks by March 2007
Ayacut developed Ayacut to beBlock Contemplated with field developed with
Sl number number ayacut channels field channels
1 I 16,833.32 7,805 9,028.32
2 II 3,814.81 1,665 2,149.81
3 III 4,745.62 1,410 3,335.62
4 IV 8,869.92 2,090 6,779.92
5 V 4,324.91 1,200 3,124.91
6 VI 4,716.66 2,087 2,629.66
7 VII 13,170.44 7,000 6,170.44
8 VIII 6,333.41 1,500 4,833.41
9 IX 10,812.88 4,600 6,212.88
10 X 24,117.75 8,500 15,617.75
11 XI 4,520.17 3,944 576.17
12 XIA 4,073.11 3,832 241.11
13 XII 2,258.44 1,759 499.44
14 XIIA 5,805.33 4,885 920.33
15 XIII 13,248.68 9,885 3,363.68
16 XIV 9,021.23 8,464 557.23
17 XV 6,165.22 6,165.22
18 XVI 11,104.31 11,104.31
Total 153,936.21 87,895.53 66,040.68
Irrigation is the top priority area of Andhra
Pradesh. Jalayagnam is a unique venture of the
state. The strategy behind successful implemen-
tation of various projects that have been identi-
fied under Jalayagnam is detailed below.
For any scheme, provision of required funds and
their timely release is utmost importance. Budget
provided vis-à-vis expenditure incurred since
2004–05 speaks for itself.
Budget Expenditure
provided incurred
Year Rs in Crores Rs in Crores
2004–05 4,200 3,331
2005–06 6,300 6,569
2006–07 10,042 9,109
2007–08 13,014
All the project works are divided into convenient
packages under category I (Rs 150 Crores and
above), category II (Rs 50 Crores and above each
package), and open category and works are en-
trusted under an EPC turnkey system. The EPC
contractors are expected to do proper planning
and execute the works, as per the program. As
many as 245 packages have been awarded so far
and work is going on as programmed.
About 6.5 lakh acres of land is required to be
acquired for all the 35 major and 17 medium
projects. So far more than 2 lakhs acres of land
have been acquired on consent award method
after conducting Gram Sabhas. The district-level
committees are delegated with powers to de-
cide the rates up to 50 percent for prevailing
rates and the state-level committee with full pow-
ers. By this method land acquisition process is
speeded up.
Various departments of the government of India
are being pursued vigorously for speedy clear-
ances of projects.
Third-party quality control agencies are entrusted
with the job of overseeing the quality of con-
struction in all these projects.
An online monitoring system developed by the
Center for Good Governance is being used by all
the supervising officers right up to the secretaries
for review of progress of works.
Monthly review meetings by the Honorable Chief
Minister are held.
The irrigation sector requires a lot of improve-
ment. The irrigation efficiency in Andhra Pradesh
is only around 35 percent. It needs to be in-
creased to at least 50 percent by tight water man-
agement. Though PIM has been in vogue since
1997, the WUAs’ performance is not showing the
anticipated results in ensuring proper water sup-
plies to tail-end areas of the systems. This is a chal-
lenge that the irrigation sector is facing, and
every possibility is being explored to have effec-
tive water management.
As per the present estimate, an amount of Rs.
72,000 crores is required to complete the ongoing
35 major and 17 medium irrigation projects for cre-
ation of additional irrigation potential of 71.17
lakh acres. The expenditure incurred since June
2004 on irrigation projects is Rs 18,000 Crores.
Irrigation potential created in the last few years
is given below.
Year Ayacut in acres
237,886
220,491
692,950
Year Proposed ayacut in acres
1,003,630
1,582,239
1,099,499
Micro Irrigation: The government issued orders
to go for micro irrigation in all lift irrigation
schemes, including the Gaaleru Nagari Sujala Sra-
vanthi Scheme. The orders are issued primarily
with a view to improve the water use efficiency
while bringing additional area under irrigation in
certain cases.
A P P E N D I X G : C O M M E N T S F R O M G O V E R N M E N T O F A N D H R A P R A D E S H
1 1 7
Attachment 2: Innovative Measures to Speed Up the System to Ground a LargeNumber of Projects
1 1 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
IEG Response to Comments
Government comment IEG response
Irrigation is definitely a powerful force for poverty reduction.
To draw conclusions on the entire irrigation development projectsis inappropriate and not justifiable.
The methodology of assessing the impact, sample size, andselection of the sample chosen by IEG has significant differencesand therefore the findings are misleading. Conclusions made onSRBC based on the sample of SRSP are inappropriate.
It may not be out of context to mention that IEG has prejudged theissues and extrapolated/generalized the small sample surveyresults across the irrigation investments.
It is very much understandable that the two projects taken up areof different nature; that is, one is basically a rehabilitation of anexisting system (SRSP) and the other one is to complete a newirrigation system for bringing new areas under irrigation. Thus,special issues concerned with these two projects andcircumstances considered by the Bank for assisting are overlookedby the study and the report is made exclusive of these facts.
The government departments will be strengthened in due coursefor maintaining efficient management of the irrigation system, soas to redress weaknesses in the activities of WUAs.
The IEG has made drastic comments on the functioning ofgovernment departments. It is felt desirable not to make an issueof this sort of comments.
[IEG’s] conclusions are not based on project-specific data . . . [so]the conclusions of the study do not adequately represent reality.
Regarding the representation of the WUAs as mostly by the elitegroups, the study conducted by the IEG is only for a limited areaand for all practical purposes the study conducted by the ORG(2005) can be taken as reliable study, as the sample taken for thesame is 214 WUAs across the state and in all three sectors ofirrigation.
IEG fully agrees with this statement, the report providing newestimates of this impact.
The report is specifically an impact evaluation of Bank supportunder AP II and AP III. Where more general statements are made,they are supported by additional evidence, notably IEG’s reviewWater Management in Agriculture: Ten Years of World BankAssistance, 1994–2004 (IEG 2006b).
The government of Andhra Pradesh’s comment appears to assumethat IEG’s analysis and finding are based solely on the IEG survey.This survey was undertaken to be able to investigate the impact ofnew irrigation. But a wide range of sources, including governmentdata and the AP III “baseline,” are used in the study.
IEG does not agree with the suggestion that the issues wereprejudged. The issue of the sample was discussed in the previouscomment.
IEG’s report does distinguish between SRSP and SRBC, both interms of outputs and effectiveness. The report is clear that SRSP isrehabilitation and SRBC new construction, and that this is a majorreason SRSP has a higher rate of return than SRBC. However, IEG’srevisions after receiving the government’s comments have madethis distinction clearer.
In revising the report, IEG has in several places indicated that thegovernment of Andhra Pradesh has introduced measures toimprove the efficiency of construction.
IEG’s comments are substantiated by reference to other studieswith similar conclusions. As noted above, attempts by thegovernment of Andhra Pradesh to improve the administration ofirrigation are noted in IEG’s revised report.
Project-specific data are used for SRSP costs and benefits andSRBC costs. Because no water had been delivered to the bulk ofthe command area of SRBC at the time the study was undertaken,benefits are estimates similar to those in the Bank’s Implementa-tion Completion Report (the government’s comments of that reportdid not question these estimates).
IEG’s conclusions regarding WUA management are not based onits survey, but on a large survey of more than 200 WUAs carriedout by cess, and a number of detailed studies. The ORG study isnow referred to in the IEG report. The general conclusion is thatWUA committees are reasonably representative, but that thepresident is not, and often runs the WUA with little consultation.
(Table continues on next page)
A P P E N D I X G : C O M M E N T S F R O M G O V E R N M E N T O F A N D H R A P R A D E S H
1 1 9
IEG Response to Comments (continued)
Government comment IEG response
The water rates could not be collected to the expected levelbecause of continuous drought. It can be observed that the watertax collections are around 100 Crores during the periods 1998–99,1999–2000, and 2000–01, as the season was good.
It is not correct that the reforms have not broken a vicious circle, as the study is limited to the SRSP Project only, and there are noWUAs in SRBC.
It is not correct to say that WUAs have only limited means toresolve water-allocation disputes, per the provisions of the Act.The farmers’ organizations have been given powers for settlementof disputes (§§26, 27 of Act).
It is not correct that there is a lack of support for WUAs, andgovernment is committed to stakeholder participation inmanagement of irrigation systems.
In the revised report IEG gives the most recently available data onwater cess collection, which remains inadequate for O&M.
As noted above, cess collection remains lower than required, andproblems in water availability remain. The revised report gives thedate of WUAs being created in SRBC.
IEG recognizes these formal powers but maintains that they areinadequate to resolve water disputes, especially those betweencommunities.
The report has been revised to reflect the commitment of thegovernment to continuing the reform process.
1 2 1
Impact evaluation seeks to distinguish between
what happened with the intervention and what
would have happened in the absence of that
intervention—that is, with versus without. The
without is also called the counterfactual. Estab-
lishing a valid counterfactual is usually done using
a control group—that is, a group similar to the
treatment (beneficiary) group other than it does
not have the intervention.
One problem encountered in establishing the
counterfactual is that of possible selection bias,
whereby there is a systematic bias between treat-
ment and control communities. There are a num-
ber of ways around this problem. There are
experimental approaches that are either ran-
domized designs or natural experiments. Where
these are not possible or applicable, then a
quasi-experimental approach can be used, such
as regression discontinuity or propensity score
matching. Both these approaches are regression
based.
If selection is on observables, then regression-
based approaches can remove selection bias. If se-
lection is based on time-invariant unobservables,
then quasi-experimental approaches can also be
used if there are both baseline and endline data
(allowing double difference estimates). Only if
selection is on time-varying unobservables is it nec-
essary to use experimental approaches to avoid
selection bias.
IEG’s approach is to stress that impact evaluation
should be rigorous and relevant. That is, IEG’s
impact studies apply rigorous approaches but
also aim to deliver policy-relevant conclusions.
Doing this usually means using a theory-based ap-
proach that analyzes the whole results chain for
an intervention.
This impact study combines a variety of data
sources to examine impact. The survey used a
pipeline approach to identify the control com-
munities, and regression analysis for most in-
dicators. Other data sources allow estimates of
single, and in some cases, double-difference
estimates.
For further information on approaches to impact
evaluation, see the two IEG booklets Impact Eval-
uation: The Experience of the Independent Eval-
uation Group of the World Bank (IEG 2006a)
and Conducting Quality Impact Evaluation
under Budget, Time and Data Constraints (Bam-
berger 2006).
Recent IEG impact evaluations include: Agricul-
tural Extension: The Kenya Experience (2000);
Books, Buildings and Learning Outcomes: An Im-
pact Evaluation of World Bank Support to Basic
Education in Ghana (2004); and Maintaining
Momentum to 2015: An Impact Evaluation of Ex-
ternal Support to Maternal and Child Health
and Nutrition in Bangladesh (2005).
APPENDIX H: IMPACT EVALUATION
1 2 3
Chapter 11. Data are from FAOSTATS. For India these figures
are 2,070 and 2,470. About 2,000 calories is the mini-
mum daily requirement and is insufficient for those in-
volved in physical labor. Because these are average
figures, many people are below this level.
2. Gross cropped area “double counts” land used for
multicropping and additional cropping seasons; the lat-
ter in particular is often made possible by irrigation.
3. These findings are supported by several different
pieces of work, such as the econometric analysis with
district data of Datt and Ravallion (1998) and Fan and
Hazell (2001). A similar approach using state-level data
is Bhattarai and Narayanamoorthy (2003). Recent
project-level evidence comes from a study by the Inter-
national Water Management Institute, which collected
data from 26 irrigation systems in six Asian countries
(Hussain 2007). For a general overview on irrigation and
poverty, see Lipton, Litchfield, and Faurès (2003). For
a summary of recent Indian evidence see Malik (2005).
4. For example, average annual lending to irrigation
in the period 1994–98 was $610 million, compared
with $419 million from 1999 to 2004.
5. It has also been observed that the more rigorous
the ex post analysis, the lower the ERR (Berkoff 2002).
6. There are two aspects to excessive water con-
sumption. One is planting crops requiring large quan-
tities of water, notably paddy. The second is applying
more water than is strictly necessary.
7. The basic charge was raised from Rs 60/acre to Rs
200/acre, with some variations.
8. The legislation also put in place project commit-
tees at the scheme level, but these have not yet been
created.
9. WUAs are just one manifestation of Janmab-
hoomi. Many thousands of women’s self-help groups
were also formed. These are discussed in a separate re-
port on the Andhra Pradesh Rural Livelihoods Program
(World Bank 2004b).
10. The government of Andhra Pradesh states that
the delay occurred to allow a review of WUA perform-
ance and for necessary changes to be instituted.
11. AP II collected no baseline data. AP III collected
“baseline” survey data only in 2002, toward the end of
the project. This survey data is drawn on in the current
study.
12. The terminology here is the evaluation con-
vention of a project or treatment group (that is, those
receiving canal irrigation) and an untreated compari-
son group. The pipeline approach uses as the com-
parison group those who are scheduled to receive
the treatment but who have not yet done so, hence
overcoming the problem of selection bias caused by
drawing the comparison group from those who will not
participate. The bias arises because the determinants
of participation may be correlated with project
outcomes.
13. In its comments, the government of Andhra
Pradesh questions drawing conclusions for the whole
project based on a survey carried out in the tail end of
SRSP. However, as the text makes clear, IEG’s findings
are based on a wide variety of data sources.
14. A natural experiment means that a comparison
group arises out of a setting without intervention from
the evaluator.
15. In project villages, these village-level averages in-
clude households without direct access to the irrigation
system, but that may have benefited from the recharg-
ing of the groundwater through seepage from the canal,
tanks filled with canal water, and irrigated plots.
Chapter 21. Wade (1982) documents the channels for this
rent seeking in detail, and the amounts involved at
that time. It continues to be the case that engineers pre-
fer to work in construction rather than O&M.
2. See Vaidyanathan (1999) for a discussion of these
issues.
ENDNOTES
3. Even the ICR, having discussed the problems
that prevent water reaching the lower reaches and the
fact that these problems have not been addressed,
produced ERR estimates assuming the whole com-
mand will receive water.
4. The government of Andhra Pradesh piloted the
WUA approach in the SRSP command prior to AP III.
The expansion of WUAs across the command was, in
the words of the appraisal report, to be a “prototype”
for public investment schemes elsewhere in the state.
However, the “big bang” approach adopted for the
creation of WUAs meant that the support through AP
III did not serve as a prototype.
5. This figure is for distributaries up to km 234 only.
Under SRSP, a further 113,000 ha (funded by the gov-
ernment) have been added, from km 234 to km 284.
6. See also Vaidyanatan 1999 (pp. 59–61) on dis-
crepancies between different data sources regarding
irrigated area.
7. Revenue estimates of cropped area are the basis
for tax collection, so underreporting produces a rent to
be shared between the landowner and revenue official.
8. This problem is not unique to Andhra Pradesh or
India. IEG’s irrigation sector review reported figures for
four projects in East Asia for which the actual cropped
area was, on average, 67 percent of that expected at ap-
praisal (IEG 2002, table 3.4).
9. A partial caveat is in order here, because seep-
age from the canal raises groundwater levels and so fa-
cilitates groundwater irrigation (boreholes and tube-
wells, as well as collection in other water-harvesting
structures).
10. The SAR correctly states that the AP II investment
costs are sunk costs that are irrelevant to the decision
of whether to proceed with AP III. But in evaluating the
ex post return to irrigation investments, all costs should
be included. Assessing the overall investment based only
on the follow-on project would allow a sleight of hand
to produce a good rate of return. There is a parallel with
the multistate Subernarekha Irrigation Project from
the same time as AP II. This project financed initial
construction costs, with most of the irrigated area to
be established after the project. The appraisal report
discounted fears that the later investments may not take
place on the grounds that the initial investment costs
would by then be sunk costs, so the additional invest-
ment would have a very good rate of return! In the
event, cost overruns and delays meant that there was
no irrigation at all under the project and the Bank de-
cided not to finance a follow-on project.
11. In fact, the government’s own investment in
the scheme began in 1984, though these costs are not
shown in this analysis, so the delay is longer and the
return lower than shown here.
12. IEG (2002, p. 48–49) cites two other Bank
projects in India and one in Nepal where construction
delays contributed to a reduction in the ERR by 4
percent.
Chapter 31. One of the best known studies of community
management of natural resources is Wade’s (1988)
analysis of irrigation management in Andhra Pradesh.
2. In major schemes, 50 percent is meant to go to
the WUA, 20 percent to the distributary committee, and
20 percent to the project committee; in medium
schemes the WUA gets 60 percent, and the distributary
committee 30 percent; in minor schemes the WUA
gets the full 90 percent. The other 10 percent goes to
local government.
3. However, most of the discontent was about other
aspects of the reforms, such as the increase in the
price of subsidized rice and the reduction in power sub-
sidies. There were public protests on these issues and
the Communist Party condemned the Chief Minister as
being a “daily wage laborer of the World Bank” (Prasad
2004).
4. The preface to an early World Bank review of the
Andhra Pradesh experience stated, “The program [is]
part of an overall vision for alleviation of poverty being
applied in many sectors by Andhra Pradesh’s state lead-
ers.” More explicitly, the Web site of the International
Network on Participatory Irrigation Management says,
“PIM can enable the poor to have greater voice in de-
cisions. Conversely, if poor people are excluded and
their interests neglected, then irrigation development
may disrupt livelihoods and increase inequity. The
poor may suffer disproportionately from irrigation per-
formance problems such as water shortage in tail-end
areas, and so may also stand to benefit more from per-
formance improvements. Experience shows that par-
ticipatory irrigation management offers important
opportunities to empower the poor in good gover-
nance and to provide benefits for the poor.”
5. If a landowner does not engage in production him-
self, then only the tenants are eligible to be members.
1 2 4
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
6. Few farm households in the sample report no ir-
rigation of any sort on any of their plots. In 2005, there
were 40 such households (of 567 cultivating house-
holds), none of which was a WUA member.
7. There are various explanations for the lack of
political momentum. One is that the Chief Minister’s
attention became diluted with other concerns, notably
new schemes for the Godvari basin. Second, the Sec-
retary who had been instrumental in ushering in the
reforms left his post. Third, the new government was
suspicious of the new institutions created under the pre-
vious Chief Minister and so did not rush to implement
the already delayed elections.
8. Bodies such as WUAs have a potentially important
role to play when water is scarce, so scarcity might be
expected to make the associations more active as farm-
ers try to secure water. But the absence of water alto-
gether means there is nothing to be allocated.
9. The wealth index is constructed from data on own-
ership of consumer durables, jewelry, and housing
quality. It is used rather than income, as it is less affected
by endogeneity.
10. The population is divided into four groups:
scheduled castes, scheduled tribes, backward castes,
and other castes. The first two categories benefit from
reservation and other rights written into the consti-
tution. Backward castes were later identified as also
being disadvantaged groups requiring special sup-
port. The remainder, other castes, are the higher caste
groups.
11. It has been argued that this has not been the gen-
eral pattern in India in the past, as there has been a neg-
ative correlation between the “irrigation ratio” (the
proportion of a household’s land that is irrigated) and
total landholdings, though that negative correlation
has weakened in recent years (Vaidyanathan 1999, p.
85). But larger farmers have a larger absolute area of
irrigated land, resulting in a marked inequality: a World
Bank study reported that small and marginal farmers
accounted for 81 percent of agricultural households at
the national level but operate only 45 percent of the ir-
rigated land (World Bank 2003, p. 10).
12. The appraisal report for AP III envisaged that
WUAs would also branch into transportation, market-
ing, and agroprocessing. These things have not hap-
pened in any areas covered by this report.
13. This case illustrates one point (that the water flow
to the tanks is meant to be limited) but raises another
question (why the Irrigation Department has not op-
posed it elsewhere, raising the specter of political in-
fluence and rent-seeking).
14. This is the same figure as that found in another
study of WUAs in Andhra Pradesh (Alsop and others
2002).
15. The government of Andhra Pradesh disputes the
assertion that WUAs are subject to elite capture, citing
the ORG report (2005) as evidence. While it is true that
the WAU committees are more representative, the key
position of president is not. The lack of consultation
by the president is supported by the ORG report.
16. Other measures of socioeconomic status—
wealth, education, and an index of political status—are
correlated with landholding, so there are problems of
multicollinearity undermining their significance. If
land is dropped, then wealth and political status be-
come significant. A similar analysis is presented in the
paper of Alsop and others (2002), who asked WUA
members only. Large landowners, the non-poor, and
those who benefited from increased water supply
were more likely to respond that the WUA achieved ef-
fective management of the irrigation system and bet-
ter water distribution.
17. The same point is made in Mott MacDonald
(2005), drawing on experience from Nepal, the Kyrgyz
Republic, as well as Andhra Pradesh. They also em-
phasize that WUAs lack the technical expertise to man-
age the canal network.
18. This finding echoes that in Jairath’s (2000) re-
view of WUAs, written during the first term of office
when there was considerable optimism about the
Andhra Pradesh experience: “The tail-end areas of
the sample distributaries continue to be deprived of
their share of water. There was no effective way of pre-
vailing over the head villages even through the medi-
ation of distributary committees.” The World Bank
appraisal document for AP III stated that the forma-
tion of WUAs would create a lobby against water waste
in higher reaches, but this does not seem to have
happened.
19. In China, for example, where the benefits of ir-
rigation are more equitably distributed than they are
in India, Communist Party officials are active in the
water-allocation process. The same author says ex-
plicitly that “smooth functioning of the system requires
that there be some high authority to enforce rules and
resolve conflicts” (Vaidyanathan 1999, p. 27).
E N D N O T E S
1 2 5
20. However, government officials still had to sanc-
tion works proposed by the WUA and Reddy and Reddy
(2005) state that there was a lot of rent seeking asso-
ciated with this process, which is to be expected given
Wade’s analysis (1982).
21. This point is made in each of Mollinga, Do-
raiswamy, and Engbersen (2001), Alsop and others
(2002), and Jairath (2000).
22. According to Wade (1982), contractors play an
important role in the system of political and adminis-
trative corruption, being responsible for “paying off ”
so most of the rents raised do not have to pass through
the hands of senior politicians. The predominance of
contractors among WUA presidents supports suspi-
cions that the situation has not improved since Wade
wrote his paper.
23. There is one documented case of farmers in
Andhra Pradesh taking it upon themselves to form
their own project-level committee and claiming, albeit
unsuccessfully, the share of the project committee
(Mollinga, Doraiswamy, and Engbersen 2001).
24. On top of the water fee there is meant to be an
“own contribution” of 15 percent from WUAs to any
works carried out. These contributions are commonly
not collected, being estimated at just 5 percent (rather
than 15 percent) of expenditures.
Chapter 41. This is assuming average farm size of 1.3 ha in SRSP
(calculated from IEG survey, the SAR assumed 1.1 ha,
but the SRSP baseline gives a higher figure of 1.4 ha),
that the irrigation proportion is two-thirds, and that 80
percent of the reported ayacut (area served by the ir-
rigation scheme) actually receives irrigation water.
2. Cropping intensity can also increase with multi-
cropping, that is, growing more than one crop on the
same plot at the same time. However, multicrop-
ping is more common on rain-fed plots than irrigated
ones.
3. See appendix table D.8 for actual and expected
yields for major crops.
4. The World Bank’s ICR for AP III makes the same
observation.
5. “Double-difference” estimates of actual yield in-
creases were calculated using the survey data at both
plot and community levels. At the plot level, the first
difference is the average change in yields for plots that
grew paddy in both 2005 and 2006 but that used canal
or tank irrigation only in the latter year. This becomes
a double difference by subtracting from this yield in-
crease the yield increase in an “untreated” control area
(see appendix C). Two control groups were used. The
first control was a panel of plots that had not changed
their irrigation source between the two years. The sec-
ond control comprised plots that were not irrigated
during either year (no canal, tank, borehole, or tube-
well). In the latter case, the treatment group included
those with newly acquired canal or tank irrigation
who did not have irrigation in 2005. Although the
second double difference is a more satisfactory defi-
nition, the sample size becomes small, so the first
method has to be used for less common crops. This
analysis is complemented by a fixed effects regression
of the panel data.
6. The staff of the Agricultural Finance Corporation
assisted in the acquisition of these data.
7. The ICR yield increase estimates for paddy are
lower than this range.
8. This analysis was also performed controlling for
soil type, but the subsample sizes for the unirrigated
plots were too small to allow a meaningful comparison.
9. If the comparison group is well chosen, then the
treatment and comparison group have the same initial
conditions, so that the ex post single difference is equal
to the double difference.
10. It might be argued that farmers will take time to
realize the full yield potential. However, the yields of
newly irrigated farmers are not notably different from
those of already irrigated farmers, suggesting that there
is little scope for “catch-up.”
11. The shortfalls reported here are very similar to
those found in a review of four projects in East Asia, with
actual yields being 40, 47, 48, and 73 percent of those
expected at appraisal (IEG 2002, table 3.3).
12. The benefit stream is net income, that is, revenue
minus costs. The percentage change in net income is
the percentage change in revenue (–50 percent) times
the ratio of revenue to net income, which is taken as
1.5. Hence, the reduction in net income is 75 percent.
13. The appraisal ERRs are those given in the SAR,
which do not include the AP II sunk costs. According
to the SAR, taking these sunk costs into account reduces
the overall ERR by just over 5 percent and that for
SRSP below Lower Manair Dam by just over 3 percent.
Hence, the ERRs with a 50 percent reduction of yields,
and allowing for sunk costs, are 9 percent for SRSP
below Lower Manair Dam and 3 percent for the whole
project.
1 2 6
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
14. Although those with new access to canal or tank
irrigation appear more diversified, this may simply in-
dicate that farmers do not immediately adjust to the less-
diversified cropping pattern typical of those who have
access to irrigation water for a longer period. Moreover,
the double-difference estimates (comparing the change
in rice shares between 2005 and 2006 for households
that have newly acquired access to canal or tank irri-
gation) show an increase in rice share of 5–17 percent
for those who did not have irrigation previously and of
5–9 percent for those previously using other forms of
irrigation.
15. See Jairath 2000 for a discussion of the reasons
for rice preference.
16. In Andhra Pradesh the rice price for consumers
is guaranteed by the government. If the producer price
remains above this consumer price, then the effect of
the price reduction is a fiscal saving for government
rather than a benefit for consumers.
17. Vaidyanathan’s (1999) review of Indian irrigation
reports estimates in the range of 20– 30 percent.
18. In the plot-level regression for cropping inten-
sity, none of the irrigation source dummies is signifi-
cant once the number of sources is included in the
regression.
19. Three-quarters of plots with access to canal and
tank irrigation acquire the water by flooding, whereas
the majority of those with tubewell or borehole irri-
gation use an electric pump (appendix table D.14).
20. The IWMI study (Hussain 2007) concluded that
the multiplier in India was between 1.22 and 3, typically
around 2.21, which is higher than, but not inconsistent
with, Hazell and Haggblade’s (1990) finding that a Rs
100 increase in agricultural incomes increases non-
farm income by Rs 64, that is, a multiplier of 1.64. Bhat-
tarai, Barker, and Narayanamoorthy (2004) suggest an
even higher multiplier of 3–4.5.
21. A shadow wage rate of 0.5 has been used in this
report (in line with the appraisal and ICR calculations),
thus assuming that half the employment is additional
or that the alternative employment was only half as pro-
ductive.
22. Calculated for cultivating households only (see
appendix table D.15).
23. This explanation is supported by two findings.
First, in 2005 only 16 percent of cultivating households
hired out labor for agricultural purposes, but in 2006
nearly two-thirds (64 percent) did so. Second, the fixed
effects regression for domestic labor has a negative
coefficient on the canal dummy but a positive one on
the (canal x area) interactive variable, showing that in
larger households canal irrigation increases use of do-
mestic labor (see appendix C for results).
24. Using the constant 2005 price value of farm out-
put, labor productivity was Rs 101 per day’s labor in 2005
and 118 in 2006 (the calculation excludes farms with
failed crops).
25. The mean wage for women rose from Rs
24.95/day to Rs 27.50/day (and the median from Rs 25
to Rs 30). The mean for men rose from Rs 48.09 to Rs
50.72 (though the median remained unchanged at Rs
50); see appendix table D.24.
26. This increase in outside employment was al-
most entirely among cultivators: average outside em-
ployment rose by 100 days a year for cultivating families,
whereas there was no significant change for nonculti-
vating households (appendix table D.15), a finding
that mirrors the drop in the use of domestic labor.
27. For example, literacy in head reaches is higher
than in tail reaches, though the difference is not great
(48.0 versus 43.5 percent—calculated from census
data). The main difference is in the distance to Waran-
gal, which is on average twice as far for end-reach vil-
lages as for those at distributary heads (appendix table
D.26).
28. There is an important caveat that the analysis did
not include some benefits. First, it did not allow for the
more severe drought-related shocks on rain-fed plots.
Second, multiplier effects are not included. Third, the
welfare impact of lower food prices is not considered.
29. The benefit inflator is 13 percent (=1/0.88, not
12 percent) because the appropriate base is the share
used for canal irrigation. The assumption here is that
the benefit per unit of water is the same regardless of
end use, which is almost certainly not so, but as good
as any other assumption in the absence of data to
quantify the benefit from alternative uses.
30. The SRBC main canal is 141 km, which is to ir-
rigate 65,000 ha, which compares to the ayacut of up
to 323,000 ha covered by the 234 km of the Kakatiya
main canal in SRSP—that is, the irrigated area per km
of main canal in SRBC is one-third of that in SRBC.
31. The SAR for AP III did consider just one combi-
nation: the two construction-related risks of cost over-
run and delayed start.
32. This calculation subtracts from the ERR base
case return the decrease caused by allowing for sunk
costs (–5.2 percent), a 25 percent reduction in yields
E N D N O T E S
1 2 7
(–2.4 percent), a 20 percent increase in costs (–2.4 per-
cent), a two-year lag in benefits (–5.0 percent), and lim-
ited diversification (–5.4 percent).
Appendix A1. The head reach is the upper 25 percent in length
and the tail the bottom 25 percent. Reaches were iden-
tified by matching a village list against a list of WUAs,
giving villages and minor canals (subdistributary num-
bers) and using a system map to classify the minors by
reach. A small percentage of villages could not be iden-
tified (spellings of the same village can vary greatly by
the source consulted and even in different rounds of
the census) and so are listed as unclassified.
2. A village can be identified by its reach on each of
the main, distributary, and minor canals; for example,
a village with the position middle, head, tail, would be
a village in the middle reach of the main canal, at the
head of the distributary and toward the tail of the
minor. The reach referred to in the sample design is that
in the distributary canal. All sample villages fall within
approximately the same reach of the main canal, which
is the tail, though this was relatively well served by
water in 2005–06, given the good monsoon. The po-
sition of the minor also matters, as shown by the fact
that we found several instances in which villages at
the head of the minor captured all the water.
Appendix E1. It can be noted in passing that the very high cost
per irrigated hectare for SRBC requires substantial
yield increases to achieve an acceptable rate of return.
2. The poverty headcount actually falls by less under
scenario 2. This is because under scenario 2 the income
gain to those in the second quartile is less, so fewer cross
the poverty line. Although the income gain for those
in the first quartile is more under this scenario, it is in-
sufficient for them to reach the poverty line.
Appendix G1. Consolidated comments are prepared by IEG on
the basis of separate sets of comments submitted to IEG.
1 2 8
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
1 2 9
Alsop, Ruth, Disa Sjoblom, Ceema Namazie, and Pawan
Patil. 2002. “Community Level User Groups in Three
World Bank–Aided Projects: Do They Perform as Ex-
pected?” Social Development Paper No. 40, World
Bank, Washington, DC.
Bamberger, Michael. 2006. Conducting Quality Impact
Evaluations under Budget, Time and Data Con-
straints. ECD Series. Washington, DC: World Bank.
Belli. P., J. Anderson, H. Barnum, J. Dixon, and T. Jee-
Peng. 2002. Economic Analysis of Investment
Operations: Analytical Tools and Practical Appli-
cations. Washington, DC: World Bank Institute.
Berkoff, Jeremy. 2002. “Economic Evaluation: Why Is
It So Often Unsatisfactory? And Why Does This
Matter?” Paper presented at the International Con-
sulting Economists Association Meeting, London,
June 19.
Bhattarai, M., Randolph Barker, and A. Narayanamoor-
thy. 2004. “Who Benefits from Irrigation Develop-
ment in India? Implications of Irrigation Multipliers
for Cost Recovery and Irrigation Financing.” Paper
presented at the International Commission on Ir-
rigation and Drainage Asian Regional Workshop,
Taipei, November 10–12.
Bhattarai, M., and A. Narayanamoorthy. 2003. “Impact
of Irrigation on Rural Poverty in India: An Aggregate
Panel Data Analysis.” Water Policy 5: 443–58.
Datt, Gaurav, and Martin Ravallion. 1998. “Why Have
Some Indian States Performed Better than Others
at Reducing Rural Poverty?” Economica 65 (257):
17–38.
Fan, Shenggen, and Peter Hazell. 2001. “Returns to
Public Investments in the Less-Favored Areas of
India and China.” American Journal of Agricul-
tural Economics 83 (5): 1217–22.
Hazell, Peter, and Steven Haggblade. 1990. “Rural-
Urban Growth Linkages in India.” Agriculture and
Rural Development Department Working Paper,
World Bank, Washington, DC.
Hussain, Intizar. 2007. “Pro-Poor Intervention Strategies
in Irrigated Agriculture in Asia: Issues, Lessons, Op-
tions and Guidelines.” Irrigation and Drainage 56
(2–3): 119–26.
I&CADD (Irrigation & Command Area Development
Department). 2004. Third Andhra Pradesh Irriga-
tion Project: Project Completion Report. Hyder-
abad: Government of Andhra Pradesh.
IEG (Independent Evaluation Group). 2006a. Impact
Evaluation—The Experience of the Independent
Evaluation Group of the World Bank. Washington,
DC: World Bank.
———. 2006b. Water Management in Agriculture:
Ten Years of World Bank Assistance, 1994–2004.
IEG Study Series. Washington, DC: World Bank.
———. 2005. Maintaining Momentum to 2015: An Im-
pact Evaluation of External Support to Maternal
and Child Health and Nutrition in Bangladesh. IEG
Study Series. Washington, DC: World Bank.
———. 2004. Books, Buildings, and Learning Out-
comes: An Impact Evaluation of World Bank Sup-
port for Basic Education in Ghana. IEG Study
Series. Washington, DC: World Bank.
———. 2002. Bridging Troubled Waters: Assessing the
World Bank Water Resources Strategy. Washington,
DC: World Bank.
———. 2000. Agricultural Extension: The Kenya Ex-
perience. An Impact Evaluation. IEG Study Series.
Washington, DC: World Bank.
———. 1998. “Recent Experiences with Involuntary Re-
settlement: Overview.” Report No. 17538, World
Bank, Washington, DC.
Jairath, Jasveen. 2000. “Participatory Irrigation Man-
agement in Andhra Pradesh: Contradictions of a
Supply Side Approach.” Paper presented at South
Asia Regional Poverty Monitoring and Evaluation
Workshop, New Delhi.
Krishna, Anirudh, Mahesh Kapila, Sharad Pathak, Ma-
hendra Prowal, Kiranpal Singh, and Virpal Singh.
BIBLIOGRAPHY
2004. “Falling into Poverty in Villages of Andhra
Pradesh: Why Poverty Avoidance Policies Are
Needed.” Economic and Political Weekly 39 (20):
3249–56.
Lipton, Michael, Julie Litchfield, and Jean-Marc Faurès.
2003. “The Effects of Irrigation on Poverty.” Water
Policy 5: 413–27.
Malik, R.P.S. 2005. “Interventions in Development and
Management of Water and How They Impact on
Poverty.” Background Paper, World Bank, Washing-
ton, DC.
Mollinga, Peter, R. Doraiswamy, and Kim Engbersen.
2001. “Participatory Irrigation Management in
Andhra Pradesh, India.” International Journal of
Water 1 (3–4): 71–90.
Mott MacDonald. 2005. Equity, Irrigation and Poverty:
How to Distribute Water to the Poor. London: U.K.
Department for International Development.
Oblitas, Keith, J. Raymond Peter, Gautam Pingle, Halla
M. Qaddumi, and Jayantha Perera. 1999. “Transfer-
ring Irrigation Management to Farmers in Andhra
Pradesh, India.” Technical Paper No. 449, World
Bank, Washington, DC.
ORG Centre for Social Research. 2005. Performance
Evaluation of Farmers’ Organizations in Andhra
Pradesh. Hyderabad: I&CADD.
Plusquellec, Hervé. 2002. How Design, Management
and Policy Affect the Performance of Irrigation
Projects. Bangkok: Food and Agriculture Organiza-
tion of the United Nations.
Prasad, R.J. Rajendra. 2004. Emergence of Telugu
Desam. Hyderabad: Masterminds Academic Press.
Raju, K.V. 2001. “Participatory Irrigation Management
in India (Andhra Pradesh).” Paper for International
E-mail Conference on Irrigation Management Trans-
fer, June–October.
Reddy, V. Ratna, and P. Prudhvikar Reddy. 2005. “How
Participatory Is Participatory Irrigation Manage-
ment? A Study of WUAs in Andhra Pradesh.” CESS
Working Paper No. 65, Center for Economic and So-
cial Studies, Hyderabad.
Sivamohan, S.V.K., and C. Scott. 2005. “Coalition Build-
ing for Participatory Irrigation Management under
Changing Water Resource Trends: Reflections on
the Reforms in Andhra Pradesh.” In Asian Irrigation
in Transition: Responding to Challenges, ed. Ganesh
Shivakoti, Douglas Vermillion, Wai-Fung Lam, and Eli-
nor Ostrom. Thousand Oaks, CA: Sage Publications.
Thakkar, Himanshu. 1999. “Assessment on Irrigation in
India.” Contributing Paper of the World Commission
on Dams, London.
Upadhyay, B. 2004. “Gender Roles and Multiple Uses
of Water in North Gujarat.” Working Paper. Inter-
national Water Management Institute, Colombo.
Vaidyanathan, A. 1999. Water Resource Management:
Institutions and Irrigation Development in India.
New Delhi: Oxford University Press.
Wade, Robert. 1988. Village Republics. Cambridge,
U.K.: Cambridge University Press.
———. 1982. “The System of Administrative and Po-
litical Corruption: Canal Irrigation in South India.”
Journal of Development Studies 18 (3): 287–327.
White, Howard. 2006. Impact Evaluation: The Expe-
rience of the Independent Evaluation Group of
the World Bank. ECD Series. Washington, DC: World
Bank.
———. 2004. “Challenges in Evaluating Development
Effectiveness.” In Evaluating Development Effec-
tiveness, Vol. 7 of World Bank Series on Evaluation
and Development, eds. George Keith Pitman, Os-
valdo N. Feinstein, and Gregory K. Ingram, 33–54.
Edison, NJ: Transaction Publishers.
World Bank. 2005. India’s Water Economy: Bracing
for a Turbulent Future. Washington, DC: World
Bank.
———. 2004a. Water Resources Sector Strategy: Strate-
gic Directions for World Bank Engagement. Wash-
ington, DC: World Bank.
———. 2004b. “Women’s Self-Help Groups in Andhra
Pradesh—Participating Poverty Alleviation in Ac-
tion.” Working Paper No. 30831, World Bank, Wash-
ington, DC.
———. 2003. “The Incidence of Canal Irrigation
Subsidies in India: A Policy Note.” South Asia Rural
Development Unit Document, World Bank, Wash-
ington, DC.
———. 1997. “India—Water Resources Management
Sector Review: Urban Water Supply and Sanitation
Report.” Irrigation Sector Report, World Bank, Wash-
ington, DC.
1 3 0
A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S
WORKING FOR A WORLD FREE OF POVERTY
The World Bank Group consists of five institutions—the International Bank for Reconstruction and Development(IBRD), the International Finance Corporation (IFC), the International Development Association (IDA), theMultilateral Investment Guarantee Agency (MIGA), and the International Centre for the Settlement of InvestmentDisputes (ICSID). Its mission is to fight poverty for lasting results and to help people help themselves and theirenvironment by providing resources, sharing knowledge, building capacity, and forging partnerships in the publicand private sectors.
THE WORLD BANK GROUP
ENHANCING DEVELOPMENT EFFECTIVENESS THROUGH EXCELLENCE AND INDEPENDENCE IN EVALUATION
The Independent Evaluation Group (IEG) is an independent, three-part unit within the World Bank Group. IEG-World Bank is charged with evaluating the activities of the IBRD (the World Bank) and IDA, IEG-IFC focuses onassessment of IFC’s work toward private sector development, and IEG-MIGA evaluates the contributions of MIGAguarantee projects and services. IEG reports directly to the Bank’s Board of Directors through the Director-General,Evaluation.
The goals of evaluation are to learn from experience, to provide an objective basis for assessing the results of theBank Group’s work, and to provide accountability in the achievement of its objectives. It also improves Bank Groupwork by identifying and disseminating the lessons learned from experience and by framing recommendations drawnfrom evaluation findings.
THE INDEPENDENT EVALUATION GROUP